├── .DS_Store
├── .gitattributes
├── .gitignore
├── .idea
    ├── change_detection.iml
    ├── dictionaries
    │   └── leon.xml
    ├── misc.xml
    ├── modules.xml
    ├── vcs.xml
    └── workspace.xml
├── LICENSE
├── README.md
├── data_loaders.py
├── datasets.py
├── losses.py
├── metrics.py
├── models.py
├── test.py
├── train.py
└── utils.py


/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/Hhhhhhao/change_detection/13b87c02166cc98d39d8be240a07abcf12893fe3/.DS_Store


--------------------------------------------------------------------------------
/.gitattributes:
--------------------------------------------------------------------------------
1 | # Auto detect text files and perform LF normalization
2 | * text=auto
3 | 


--------------------------------------------------------------------------------
/.gitignore:
--------------------------------------------------------------------------------
  1 | #data folder
  2 | data/*
  3 | models/*
  4 | 
  5 | # Byte-compiled / optimized / DLL files
  6 | __pycache__/
  7 | *.py[cod]
  8 | *$py.class
  9 | 
 10 | # C extensions
 11 | *.so
 12 | 
 13 | # Distribution / packaging
 14 | .Python
 15 | build/
 16 | develop-eggs/
 17 | dist/
 18 | downloads/
 19 | eggs/
 20 | .eggs/
 21 | lib/
 22 | lib64/
 23 | parts/
 24 | sdist/
 25 | var/
 26 | wheels/
 27 | pip-wheel-metadata/
 28 | share/python-wheels/
 29 | *.egg-info/
 30 | .installed.cfg
 31 | *.egg
 32 | MANIFEST
 33 | 
 34 | # PyInstaller
 35 | #  Usually these files are written by a python script from a template
 36 | #  before PyInstaller builds the exe, so as to inject date/other infos into it.
 37 | *.manifest
 38 | *.spec
 39 | 
 40 | # Installer logs
 41 | pip-log.txt
 42 | pip-delete-this-directory.txt
 43 | 
 44 | # Unit test / coverage reports
 45 | htmlcov/
 46 | .tox/
 47 | .nox/
 48 | .coverage
 49 | .coverage.*
 50 | .cache
 51 | nosetests.xml
 52 | coverage.xml
 53 | *.cover
 54 | .hypothesis/
 55 | .pytest_cache/
 56 | 
 57 | # Translations
 58 | *.mo
 59 | *.pot
 60 | 
 61 | # Django stuff:
 62 | *.log
 63 | local_settings.py
 64 | db.sqlite3
 65 | 
 66 | # Flask stuff:
 67 | instance/
 68 | .webassets-cache
 69 | 
 70 | # Scrapy stuff:
 71 | .scrapy
 72 | 
 73 | # Sphinx documentation
 74 | docs/_build/
 75 | 
 76 | # PyBuilder
 77 | target/
 78 | 
 79 | # Jupyter Notebook
 80 | .ipynb_checkpoints
 81 | 
 82 | # IPython
 83 | profile_default/
 84 | ipython_config.py
 85 | 
 86 | # pyenv
 87 | .python-version
 88 | 
 89 | # pipenv
 90 | #   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
 91 | #   However, in case of collaboration, if having platform-specific dependencies or dependencies
 92 | #   having no cross-platform support, pipenv may install dependencies that don't work, or not
 93 | #   install all needed dependencies.
 94 | #Pipfile.lock
 95 | 
 96 | # celery beat schedule file
 97 | celerybeat-schedule
 98 | 
 99 | # SageMath parsed files
100 | *.sage.py
101 | 
102 | # Environments
103 | .env
104 | .venv
105 | env/
106 | venv/
107 | ENV/
108 | env.bak/
109 | venv.bak/
110 | 
111 | # Spyder project settings
112 | .spyderproject
113 | .spyproject
114 | 
115 | # Rope project settings
116 | .ropeproject
117 | 
118 | # mkdocs documentation
119 | /site
120 | 
121 | # mypy
122 | .mypy_cache/
123 | .dmypy.json
124 | dmypy.json
125 | 
126 | # Pyre type checker
127 | .pyre/
128 | 


--------------------------------------------------------------------------------
/.idea/change_detection.iml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8"?>
 2 | <module type="PYTHON_MODULE" version="4">
 3 |   <component name="NewModuleRootManager">
 4 |     <content url="file://$MODULE_DIR$" />
 5 |     <orderEntry type="jdk" jdkName="Python 3.6 (BEng) (1)" jdkType="Python SDK" />
 6 |     <orderEntry type="sourceFolder" forTests="false" />
 7 |   </component>
 8 |   <component name="TestRunnerService">
 9 |     <option name="PROJECT_TEST_RUNNER" value="Unittests" />
10 |   </component>
11 | </module>


--------------------------------------------------------------------------------
/.idea/dictionaries/leon.xml:
--------------------------------------------------------------------------------
1 | <component name="ProjectDictionaryState">
2 |   <dictionary name="leon" />
3 | </component>


--------------------------------------------------------------------------------
/.idea/misc.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="JavaScriptSettings">
4 |     <option name="languageLevel" value="ES6" />
5 |   </component>
6 |   <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.6 (BEng) (1)" project-jdk-type="Python SDK" />
7 | </project>


--------------------------------------------------------------------------------
/.idea/modules.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="ProjectModuleManager">
4 |     <modules>
5 |       <module fileurl="file://$PROJECT_DIR$/.idea/change_detection.iml" filepath="$PROJECT_DIR$/.idea/change_detection.iml" />
6 |     </modules>
7 |   </component>
8 | </project>


--------------------------------------------------------------------------------
/.idea/vcs.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="VcsDirectoryMappings">
4 |     <mapping directory="$PROJECT_DIR$" vcs="Git" />
5 |   </component>
6 | </project>


--------------------------------------------------------------------------------
/.idea/workspace.xml:
--------------------------------------------------------------------------------
  1 | <?xml version="1.0" encoding="UTF-8"?>
  2 | <project version="4">
  3 |   <component name="ChangeListManager">
  4 |     <list default="true" id="84690527-123e-43c2-949e-0f74a9d1ff1c" name="Default" comment="">
  5 |       <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
  6 |       <change beforePath="$PROJECT_DIR$/data_loaders.py" beforeDir="false" afterPath="$PROJECT_DIR$/data_loaders.py" afterDir="false" />
  7 |       <change beforePath="$PROJECT_DIR$/train.py" beforeDir="false" afterPath="$PROJECT_DIR$/train.py" afterDir="false" />
  8 |       <change beforePath="$PROJECT_DIR$/utils.py" beforeDir="false" afterPath="$PROJECT_DIR$/utils.py" afterDir="false" />
  9 |     </list>
 10 |     <option name="EXCLUDED_CONVERTED_TO_IGNORED" value="true" />
 11 |     <option name="TRACKING_ENABLED" value="true" />
 12 |     <option name="SHOW_DIALOG" value="false" />
 13 |     <option name="HIGHLIGHT_CONFLICTS" value="true" />
 14 |     <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
 15 |     <option name="LAST_RESOLUTION" value="IGNORE" />
 16 |   </component>
 17 |   <component name="CoverageDataManager">
 18 |     <SUITE FILE_PATH="coverage/change_detection$models.coverage" NAME="models Coverage Results" MODIFIED="1566998753305" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
 19 |     <SUITE FILE_PATH="coverage/change_detection$train.coverage" NAME="train Coverage Results" MODIFIED="1567394863293" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
 20 |     <SUITE FILE_PATH="coverage/change_detection$data_loaders.coverage" NAME="data_loaders Coverage Results" MODIFIED="1567395554528" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
 21 |     <SUITE FILE_PATH="coverage/change_detection$datasets.coverage" NAME="datasets Coverage Results" MODIFIED="1567259669080" SOURCE_PROVIDER="com.intellij.coverage.DefaultCoverageFileProvider" RUNNER="coverage.py" COVERAGE_BY_TEST_ENABLED="true" COVERAGE_TRACING_ENABLED="false" WORKING_DIRECTORY="$PROJECT_DIR$" />
 22 |   </component>
 23 |   <component name="FileEditorManager">
 24 |     <leaf SIDE_TABS_SIZE_LIMIT_KEY="300">
 25 |       <file leaf-file-name="models.py" pinned="false" current-in-tab="false">
 26 |         <entry file="file://$PROJECT_DIR$/models.py">
 27 |           <provider selected="true" editor-type-id="text-editor">
 28 |             <state relative-caret-position="2071">
 29 |               <caret line="125" column="79" selection-start-line="125" selection-start-column="79" selection-end-line="125" selection-end-column="79" />
 30 |               <folding>
 31 |                 <element signature="e#0#21#0" expanded="true" />
 32 |               </folding>
 33 |             </state>
 34 |           </provider>
 35 |         </entry>
 36 |       </file>
 37 |       <file leaf-file-name="losses.py" pinned="false" current-in-tab="false">
 38 |         <entry file="file://$PROJECT_DIR$/losses.py">
 39 |           <provider selected="true" editor-type-id="text-editor">
 40 |             <state relative-caret-position="513">
 41 |               <caret line="66" column="19" selection-start-line="66" selection-start-column="19" selection-end-line="66" selection-end-column="19" />
 42 |               <folding>
 43 |                 <element signature="e#0#12#0" expanded="true" />
 44 |               </folding>
 45 |             </state>
 46 |           </provider>
 47 |         </entry>
 48 |       </file>
 49 |       <file leaf-file-name="metrics.py" pinned="false" current-in-tab="false">
 50 |         <entry file="file://$PROJECT_DIR$/metrics.py">
 51 |           <provider selected="true" editor-type-id="text-editor">
 52 |             <state relative-caret-position="1960">
 53 |               <caret line="98" column="17" selection-start-line="98" selection-start-column="17" selection-end-line="98" selection-end-column="17" />
 54 |               <folding>
 55 |                 <element signature="e#0#18#0" expanded="true" />
 56 |               </folding>
 57 |             </state>
 58 |           </provider>
 59 |         </entry>
 60 |       </file>
 61 |       <file leaf-file-name="train.py" pinned="false" current-in-tab="false">
 62 |         <entry file="file://$PROJECT_DIR$/train.py">
 63 |           <provider selected="true" editor-type-id="text-editor">
 64 |             <state relative-caret-position="553">
 65 |               <caret line="112" column="24" selection-start-line="112" selection-start-column="24" selection-end-line="112" selection-end-column="24" />
 66 |               <folding>
 67 |                 <element signature="e#0#9#0" expanded="true" />
 68 |               </folding>
 69 |             </state>
 70 |           </provider>
 71 |         </entry>
 72 |       </file>
 73 |       <file leaf-file-name="test.py" pinned="false" current-in-tab="false">
 74 |         <entry file="file://$PROJECT_DIR$/test.py">
 75 |           <provider selected="true" editor-type-id="text-editor">
 76 |             <state relative-caret-position="420">
 77 |               <caret line="21" selection-start-line="21" selection-end-line="21" />
 78 |               <folding>
 79 |                 <element signature="e#0#9#0" expanded="true" />
 80 |               </folding>
 81 |             </state>
 82 |           </provider>
 83 |         </entry>
 84 |       </file>
 85 |       <file leaf-file-name="datasets.py" pinned="false" current-in-tab="false">
 86 |         <entry file="file://$PROJECT_DIR$/datasets.py">
 87 |           <provider selected="true" editor-type-id="text-editor">
 88 |             <state relative-caret-position="280">
 89 |               <caret line="14" selection-start-line="14" selection-end-line="14" />
 90 |               <folding>
 91 |                 <element signature="e#0#9#0" expanded="true" />
 92 |               </folding>
 93 |             </state>
 94 |           </provider>
 95 |         </entry>
 96 |       </file>
 97 |       <file leaf-file-name="data_loaders.py" pinned="false" current-in-tab="true">
 98 |         <entry file="file://$PROJECT_DIR$/data_loaders.py">
 99 |           <provider selected="true" editor-type-id="text-editor">
100 |             <state relative-caret-position="153">
101 |               <caret line="8" selection-start-line="8" selection-end-line="62" selection-end-column="40" />
102 |               <folding>
103 |                 <element signature="e#0#18#0" expanded="true" />
104 |               </folding>
105 |             </state>
106 |           </provider>
107 |         </entry>
108 |       </file>
109 |       <file leaf-file-name="utils.py" pinned="false" current-in-tab="false">
110 |         <entry file="file://$PROJECT_DIR$/utils.py">
111 |           <provider selected="true" editor-type-id="text-editor">
112 |             <state relative-caret-position="320">
113 |               <caret line="16" selection-start-line="16" selection-end-line="16" />
114 |               <folding>
115 |                 <element signature="e#0#11#0" expanded="true" />
116 |               </folding>
117 |             </state>
118 |           </provider>
119 |         </entry>
120 |       </file>
121 |     </leaf>
122 |   </component>
123 |   <component name="FileTemplateManagerImpl">
124 |     <option name="RECENT_TEMPLATES">
125 |       <list>
126 |         <option value="Python Script" />
127 |       </list>
128 |     </option>
129 |   </component>
130 |   <component name="FindInProjectRecents">
131 |     <findStrings>
132 |       <find>img_size</find>
133 |       <find>iter</find>
134 |       <find>inputs, targets = next(iter(data_loader))</find>
135 |       <find>output_</find>
136 |     </findStrings>
137 |   </component>
138 |   <component name="Git.Settings">
139 |     <option name="RECENT_GIT_ROOT_PATH" value="$PROJECT_DIR$" />
140 |   </component>
141 |   <component name="IdeDocumentHistory">
142 |     <option name="CHANGED_PATHS">
143 |       <list>
144 |         <option value="$PROJECT_DIR$/MODELS.py" />
145 |         <option value="$PROJECT_DIR$/metrics.py" />
146 |         <option value="$PROJECT_DIR$/data_loader.py" />
147 |         <option value="$PROJECT_DIR$/test.py" />
148 |         <option value="$PROJECT_DIR$/log.txt" />
149 |         <option value="$PROJECT_DIR$/datasets.py" />
150 |         <option value="$PROJECT_DIR$/models.py" />
151 |         <option value="$PROJECT_DIR$/losses.py" />
152 |         <option value="$PROJECT_DIR$/train.py" />
153 |         <option value="$PROJECT_DIR$/utils.py" />
154 |         <option value="$PROJECT_DIR$/data_loaders.py" />
155 |       </list>
156 |     </option>
157 |   </component>
158 |   <component name="JsBuildToolGruntFileManager" detection-done="true" sorting="DEFINITION_ORDER" />
159 |   <component name="JsBuildToolPackageJson" detection-done="true" sorting="DEFINITION_ORDER" />
160 |   <component name="JsGulpfileManager">
161 |     <detection-done>true</detection-done>
162 |     <sorting>DEFINITION_ORDER</sorting>
163 |   </component>
164 |   <component name="NodePackageJsonFileManager">
165 |     <packageJsonPaths />
166 |   </component>
167 |   <component name="ProjectFrameBounds">
168 |     <option name="x" value="1489" />
169 |     <option name="y" value="38" />
170 |     <option name="width" value="1871" />
171 |     <option name="height" value="1162" />
172 |   </component>
173 |   <component name="ProjectView">
174 |     <navigator proportions="" version="1">
175 |       <foldersAlwaysOnTop value="true" />
176 |     </navigator>
177 |     <panes>
178 |       <pane id="ProjectPane">
179 |         <subPane>
180 |           <expand>
181 |             <path>
182 |               <item name="change_detection" type="b2602c69:ProjectViewProjectNode" />
183 |               <item name="change_detection" type="462c0819:PsiDirectoryNode" />
184 |             </path>
185 |           </expand>
186 |           <select />
187 |         </subPane>
188 |       </pane>
189 |       <pane id="Scope" />
190 |     </panes>
191 |   </component>
192 |   <component name="PropertiesComponent">
193 |     <property name="WebServerToolWindowFactoryState" value="false" />
194 |     <property name="last_opened_file_path" value="$PROJECT_DIR$" />
195 |     <property name="nodejs_interpreter_path.stuck_in_default_project" value="undefined stuck path" />
196 |     <property name="nodejs_npm_path_reset_for_default_project" value="true" />
197 |     <property name="settings.editor.selected.configurable" value="com.jetbrains.python.configuration.PyActiveSdkModuleConfigurable" />
198 |   </component>
199 |   <component name="RunDashboard">
200 |     <option name="ruleStates">
201 |       <list>
202 |         <RuleState>
203 |           <option name="name" value="ConfigurationTypeDashboardGroupingRule" />
204 |         </RuleState>
205 |         <RuleState>
206 |           <option name="name" value="StatusDashboardGroupingRule" />
207 |         </RuleState>
208 |       </list>
209 |     </option>
210 |   </component>
211 |   <component name="RunManager" selected="Python.data_loaders">
212 |     <configuration name="data_loaders" type="PythonConfigurationType" factoryName="Python" temporary="true">
213 |       <module name="change_detection" />
214 |       <option name="INTERPRETER_OPTIONS" value="" />
215 |       <option name="PARENT_ENVS" value="true" />
216 |       <envs>
217 |         <env name="PYTHONUNBUFFERED" value="1" />
218 |       </envs>
219 |       <option name="SDK_HOME" value="" />
220 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
221 |       <option name="IS_MODULE_SDK" value="true" />
222 |       <option name="ADD_CONTENT_ROOTS" value="true" />
223 |       <option name="ADD_SOURCE_ROOTS" value="true" />
224 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
225 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/data_loaders.py" />
226 |       <option name="PARAMETERS" value="" />
227 |       <option name="SHOW_COMMAND_LINE" value="false" />
228 |       <option name="EMULATE_TERMINAL" value="false" />
229 |       <option name="MODULE_MODE" value="false" />
230 |     </configuration>
231 |     <configuration name="datasets" type="PythonConfigurationType" factoryName="Python" temporary="true">
232 |       <module name="change_detection" />
233 |       <option name="INTERPRETER_OPTIONS" value="" />
234 |       <option name="PARENT_ENVS" value="true" />
235 |       <envs>
236 |         <env name="PYTHONUNBUFFERED" value="1" />
237 |       </envs>
238 |       <option name="SDK_HOME" value="" />
239 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
240 |       <option name="IS_MODULE_SDK" value="true" />
241 |       <option name="ADD_CONTENT_ROOTS" value="true" />
242 |       <option name="ADD_SOURCE_ROOTS" value="true" />
243 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
244 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/datasets.py" />
245 |       <option name="PARAMETERS" value="" />
246 |       <option name="SHOW_COMMAND_LINE" value="false" />
247 |       <option name="EMULATE_TERMINAL" value="false" />
248 |       <option name="MODULE_MODE" value="false" />
249 |     </configuration>
250 |     <configuration name="models" type="PythonConfigurationType" factoryName="Python" temporary="true">
251 |       <module name="change_detection" />
252 |       <option name="INTERPRETER_OPTIONS" value="" />
253 |       <option name="PARENT_ENVS" value="true" />
254 |       <envs>
255 |         <env name="PYTHONUNBUFFERED" value="1" />
256 |       </envs>
257 |       <option name="SDK_HOME" value="" />
258 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
259 |       <option name="IS_MODULE_SDK" value="true" />
260 |       <option name="ADD_CONTENT_ROOTS" value="true" />
261 |       <option name="ADD_SOURCE_ROOTS" value="true" />
262 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
263 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/models.py" />
264 |       <option name="PARAMETERS" value="" />
265 |       <option name="SHOW_COMMAND_LINE" value="false" />
266 |       <option name="EMULATE_TERMINAL" value="false" />
267 |       <option name="MODULE_MODE" value="false" />
268 |     </configuration>
269 |     <configuration name="train" type="PythonConfigurationType" factoryName="Python" temporary="true">
270 |       <module name="change_detection" />
271 |       <option name="INTERPRETER_OPTIONS" value="" />
272 |       <option name="PARENT_ENVS" value="true" />
273 |       <envs>
274 |         <env name="PYTHONUNBUFFERED" value="1" />
275 |       </envs>
276 |       <option name="SDK_HOME" value="" />
277 |       <option name="WORKING_DIRECTORY" value="$PROJECT_DIR$" />
278 |       <option name="IS_MODULE_SDK" value="true" />
279 |       <option name="ADD_CONTENT_ROOTS" value="true" />
280 |       <option name="ADD_SOURCE_ROOTS" value="true" />
281 |       <EXTENSION ID="PythonCoverageRunConfigurationExtension" runner="coverage.py" />
282 |       <option name="SCRIPT_NAME" value="$PROJECT_DIR$/train.py" />
283 |       <option name="PARAMETERS" value="" />
284 |       <option name="SHOW_COMMAND_LINE" value="false" />
285 |       <option name="EMULATE_TERMINAL" value="false" />
286 |       <option name="MODULE_MODE" value="false" />
287 |     </configuration>
288 |     <list>
289 |       <item itemvalue="Python.models" />
290 |       <item itemvalue="Python.datasets" />
291 |       <item itemvalue="Python.train" />
292 |       <item itemvalue="Python.data_loaders" />
293 |     </list>
294 |     <recent_temporary>
295 |       <list>
296 |         <item itemvalue="Python.data_loaders" />
297 |         <item itemvalue="Python.train" />
298 |         <item itemvalue="Python.datasets" />
299 |         <item itemvalue="Python.models" />
300 |       </list>
301 |     </recent_temporary>
302 |   </component>
303 |   <component name="SvnConfiguration">
304 |     <configuration />
305 |   </component>
306 |   <component name="TaskManager">
307 |     <task active="true" id="Default" summary="Default task">
308 |       <changelist id="84690527-123e-43c2-949e-0f74a9d1ff1c" name="Default" comment="" />
309 |       <created>1566984986195</created>
310 |       <option name="number" value="Default" />
311 |       <option name="presentableId" value="Default" />
312 |       <updated>1566984986195</updated>
313 |     </task>
314 |     <servers />
315 |   </component>
316 |   <component name="ToolWindowManager">
317 |     <frame x="1489" y="38" width="1871" height="1162" extended-state="0" />
318 |     <editor active="true" />
319 |     <layout>
320 |       <window_info anchor="bottom" id="TODO" order="6" />
321 |       <window_info anchor="bottom" id="Event Log" order="7" side_tool="true" />
322 |       <window_info anchor="bottom" id="Database Changes" order="7" show_stripe_button="false" />
323 |       <window_info anchor="bottom" id="Run" order="2" weight="0.32917467" />
324 |       <window_info anchor="bottom" id="Version Control" order="7" />
325 |       <window_info anchor="bottom" id="Python Console" order="7" />
326 |       <window_info anchor="bottom" id="Terminal" order="7" />
327 |       <window_info active="true" content_ui="combo" id="Project" order="0" visible="true" weight="0.14488792" />
328 |       <window_info anchor="bottom" id="Docker" order="7" show_stripe_button="false" />
329 |       <window_info anchor="right" id="Database" order="3" />
330 |       <window_info anchor="right" id="SciView" order="3" weight="0.32968834" />
331 |       <window_info id="Structure" order="1" side_tool="true" weight="0.25" />
332 |       <window_info anchor="bottom" id="Debug" order="3" weight="0.32821497" />
333 |       <window_info id="Favorites" order="2" side_tool="true" />
334 |       <window_info anchor="right" content_ui="combo" id="Hierarchy" order="2" weight="0.25" />
335 |       <window_info anchor="bottom" id="Inspection" order="5" weight="0.4" />
336 |       <window_info anchor="right" id="Commander" internal_type="SLIDING" order="0" type="SLIDING" weight="0.4" />
337 |       <window_info anchor="right" id="Ant Build" order="1" weight="0.25" />
338 |       <window_info anchor="bottom" id="Message" order="0" />
339 |       <window_info anchor="bottom" id="Cvs" order="4" weight="0.25" />
340 |       <window_info anchor="bottom" id="Find" order="1" />
341 |     </layout>
342 |   </component>
343 |   <component name="TypeScriptGeneratedFilesManager">
344 |     <option name="version" value="1" />
345 |   </component>
346 |   <component name="VcsContentAnnotationSettings">
347 |     <option name="myLimit" value="2678400000" />
348 |   </component>
349 |   <component name="XDebuggerManager">
350 |     <breakpoint-manager>
351 |       <breakpoints>
352 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
353 |           <url>file://$PROJECT_DIR$/utils.py</url>
354 |           <line>100</line>
355 |           <option name="timeStamp" value="43" />
356 |         </line-breakpoint>
357 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
358 |           <url>file://$PROJECT_DIR$/datasets.py</url>
359 |           <line>54</line>
360 |           <option name="timeStamp" value="47" />
361 |         </line-breakpoint>
362 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
363 |           <url>file://$PROJECT_DIR$/datasets.py</url>
364 |           <line>57</line>
365 |           <option name="timeStamp" value="48" />
366 |         </line-breakpoint>
367 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
368 |           <url>file://$PROJECT_DIR$/datasets.py</url>
369 |           <line>47</line>
370 |           <option name="timeStamp" value="49" />
371 |         </line-breakpoint>
372 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
373 |           <url>file://$PROJECT_DIR$/datasets.py</url>
374 |           <line>52</line>
375 |           <option name="timeStamp" value="50" />
376 |         </line-breakpoint>
377 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
378 |           <url>file://$PROJECT_DIR$/utils.py</url>
379 |           <line>16</line>
380 |           <option name="timeStamp" value="82" />
381 |         </line-breakpoint>
382 |         <line-breakpoint enabled="true" suspend="THREAD" type="python-line">
383 |           <url>file://$PROJECT_DIR$/data_loaders.py</url>
384 |           <line>53</line>
385 |           <option name="timeStamp" value="84" />
386 |         </line-breakpoint>
387 |       </breakpoints>
388 |       <option name="time" value="90" />
389 |     </breakpoint-manager>
390 |   </component>
391 |   <component name="editorHistoryManager">
392 |     <entry file="file://$PROJECT_DIR$/losses.py">
393 |       <provider selected="true" editor-type-id="text-editor">
394 |         <state relative-caret-position="360">
395 |           <caret line="18" column="35" selection-start-line="18" selection-start-column="35" selection-end-line="18" selection-end-column="35" />
396 |           <folding>
397 |             <element signature="e#0#12#0" expanded="true" />
398 |           </folding>
399 |         </state>
400 |       </provider>
401 |     </entry>
402 |     <entry file="file://$PROJECT_DIR$/test.py">
403 |       <provider selected="true" editor-type-id="text-editor">
404 |         <state relative-caret-position="420">
405 |           <caret line="21" selection-start-line="21" selection-end-line="21" />
406 |           <folding>
407 |             <element signature="e#0#9#0" expanded="true" />
408 |           </folding>
409 |         </state>
410 |       </provider>
411 |     </entry>
412 |     <entry file="file://$PROJECT_DIR$/datasets.py">
413 |       <provider selected="true" editor-type-id="text-editor">
414 |         <state relative-caret-position="1140">
415 |           <caret line="57" selection-start-line="57" selection-end-line="57" />
416 |           <folding>
417 |             <element signature="e#0#9#0" expanded="true" />
418 |           </folding>
419 |         </state>
420 |       </provider>
421 |     </entry>
422 |     <entry file="file://$PROJECT_DIR$/data_loaders.py">
423 |       <provider selected="true" editor-type-id="text-editor">
424 |         <state relative-caret-position="580">
425 |           <caret line="29" column="35" selection-start-line="29" selection-start-column="35" selection-end-line="29" selection-end-column="35" />
426 |         </state>
427 |       </provider>
428 |     </entry>
429 |     <entry file="file://$PROJECT_DIR$/utils.py">
430 |       <provider selected="true" editor-type-id="text-editor">
431 |         <state relative-caret-position="900">
432 |           <caret line="45" column="38" selection-start-line="45" selection-start-column="38" selection-end-line="45" selection-end-column="38" />
433 |           <folding>
434 |             <element signature="e#0#11#0" expanded="true" />
435 |           </folding>
436 |         </state>
437 |       </provider>
438 |     </entry>
439 |     <entry file="file://$PROJECT_DIR$/train.py">
440 |       <provider selected="true" editor-type-id="text-editor">
441 |         <state relative-caret-position="3900">
442 |           <caret line="195" column="42" lean-forward="true" selection-start-line="195" selection-start-column="42" selection-end-line="195" selection-end-column="42" />
443 |           <folding>
444 |             <element signature="e#0#9#0" expanded="true" />
445 |           </folding>
446 |         </state>
447 |       </provider>
448 |     </entry>
449 |     <entry file="file://$PROJECT_DIR$/models.py">
450 |       <provider selected="true" editor-type-id="text-editor">
451 |         <state relative-caret-position="2600">
452 |           <caret line="130" column="21" selection-start-line="130" selection-start-column="21" selection-end-line="130" selection-end-column="21" />
453 |           <folding>
454 |             <element signature="e#0#21#0" expanded="true" />
455 |           </folding>
456 |         </state>
457 |       </provider>
458 |     </entry>
459 |     <entry file="file://$PROJECT_DIR$/losses.py">
460 |       <provider selected="true" editor-type-id="text-editor">
461 |         <state relative-caret-position="360">
462 |           <caret line="18" column="35" selection-start-line="18" selection-start-column="35" selection-end-line="18" selection-end-column="35" />
463 |           <folding>
464 |             <element signature="e#0#12#0" expanded="true" />
465 |           </folding>
466 |         </state>
467 |       </provider>
468 |     </entry>
469 |     <entry file="file://$PROJECT_DIR$/test.py">
470 |       <provider selected="true" editor-type-id="text-editor">
471 |         <state relative-caret-position="420">
472 |           <caret line="21" selection-start-line="21" selection-end-line="21" />
473 |           <folding>
474 |             <element signature="e#0#9#0" expanded="true" />
475 |           </folding>
476 |         </state>
477 |       </provider>
478 |     </entry>
479 |     <entry file="file://$PROJECT_DIR$/datasets.py">
480 |       <provider selected="true" editor-type-id="text-editor">
481 |         <state relative-caret-position="740">
482 |           <caret line="37" column="40" selection-start-line="37" selection-start-column="40" selection-end-line="37" selection-end-column="40" />
483 |           <folding>
484 |             <element signature="e#0#9#0" expanded="true" />
485 |           </folding>
486 |         </state>
487 |       </provider>
488 |     </entry>
489 |     <entry file="file://$PROJECT_DIR$/data_loaders.py">
490 |       <provider selected="true" editor-type-id="text-editor">
491 |         <state relative-caret-position="620">
492 |           <caret line="31" selection-start-line="31" selection-end-line="31" />
493 |         </state>
494 |       </provider>
495 |     </entry>
496 |     <entry file="file://$PROJECT_DIR$/utils.py">
497 |       <provider selected="true" editor-type-id="text-editor">
498 |         <state relative-caret-position="800">
499 |           <caret line="40" column="39" selection-start-line="40" selection-start-column="39" selection-end-line="40" selection-end-column="39" />
500 |           <folding>
501 |             <element signature="e#0#11#0" expanded="true" />
502 |           </folding>
503 |         </state>
504 |       </provider>
505 |     </entry>
506 |     <entry file="file://$PROJECT_DIR$/train.py">
507 |       <provider selected="true" editor-type-id="text-editor">
508 |         <state relative-caret-position="120">
509 |           <caret line="6" column="36" selection-start-line="6" selection-start-column="36" selection-end-line="6" selection-end-column="36" />
510 |           <folding>
511 |             <element signature="e#0#9#0" expanded="true" />
512 |           </folding>
513 |         </state>
514 |       </provider>
515 |     </entry>
516 |     <entry file="file://$PROJECT_DIR$/models.py">
517 |       <provider selected="true" editor-type-id="text-editor">
518 |         <state relative-caret-position="2600">
519 |           <caret line="130" column="21" selection-start-line="130" selection-start-column="21" selection-end-line="130" selection-end-column="21" />
520 |           <folding>
521 |             <element signature="e#0#21#0" expanded="true" />
522 |           </folding>
523 |         </state>
524 |       </provider>
525 |     </entry>
526 |     <entry file="file://$PROJECT_DIR$/losses.py">
527 |       <provider selected="true" editor-type-id="text-editor">
528 |         <state relative-caret-position="360">
529 |           <caret line="18" column="35" selection-start-line="18" selection-start-column="35" selection-end-line="18" selection-end-column="35" />
530 |           <folding>
531 |             <element signature="e#0#12#0" expanded="true" />
532 |           </folding>
533 |         </state>
534 |       </provider>
535 |     </entry>
536 |     <entry file="file://$PROJECT_DIR$/test.py">
537 |       <provider selected="true" editor-type-id="text-editor">
538 |         <state relative-caret-position="420">
539 |           <caret line="21" lean-forward="true" selection-start-line="21" selection-end-line="21" />
540 |           <folding>
541 |             <element signature="e#0#9#0" expanded="true" />
542 |           </folding>
543 |         </state>
544 |       </provider>
545 |     </entry>
546 |     <entry file="file://$PROJECT_DIR$/datasets.py">
547 |       <provider selected="true" editor-type-id="text-editor">
548 |         <state relative-caret-position="740">
549 |           <caret line="37" column="40" selection-start-line="37" selection-start-column="40" selection-end-line="37" selection-end-column="40" />
550 |           <folding>
551 |             <element signature="e#0#9#0" expanded="true" />
552 |           </folding>
553 |         </state>
554 |       </provider>
555 |     </entry>
556 |     <entry file="file://$PROJECT_DIR$/data_loaders.py">
557 |       <provider selected="true" editor-type-id="text-editor">
558 |         <state relative-caret-position="620">
559 |           <caret line="31" selection-start-line="31" selection-end-line="31" />
560 |         </state>
561 |       </provider>
562 |     </entry>
563 |     <entry file="file://$PROJECT_DIR$/utils.py">
564 |       <provider selected="true" editor-type-id="text-editor">
565 |         <state relative-caret-position="800">
566 |           <caret line="40" column="39" selection-start-line="40" selection-start-column="39" selection-end-line="40" selection-end-column="39" />
567 |           <folding>
568 |             <element signature="e#0#11#0" expanded="true" />
569 |           </folding>
570 |         </state>
571 |       </provider>
572 |     </entry>
573 |     <entry file="file://$PROJECT_DIR$/train.py">
574 |       <provider selected="true" editor-type-id="text-editor">
575 |         <state relative-caret-position="120">
576 |           <caret line="6" column="36" selection-start-line="6" selection-start-column="36" selection-end-line="6" selection-end-column="36" />
577 |         </state>
578 |       </provider>
579 |     </entry>
580 |     <entry file="file://$PROJECT_DIR$/models.py">
581 |       <provider selected="true" editor-type-id="text-editor">
582 |         <state relative-caret-position="2600">
583 |           <caret line="130" column="21" selection-start-line="130" selection-start-column="21" selection-end-line="130" selection-end-column="21" />
584 |           <folding>
585 |             <element signature="e#0#21#0" expanded="true" />
586 |           </folding>
587 |         </state>
588 |       </provider>
589 |     </entry>
590 |     <entry file="file://$PROJECT_DIR$/losses.py">
591 |       <provider selected="true" editor-type-id="text-editor">
592 |         <state relative-caret-position="360">
593 |           <caret line="18" column="35" selection-start-line="18" selection-start-column="35" selection-end-line="18" selection-end-column="35" />
594 |           <folding>
595 |             <element signature="e#0#12#0" expanded="true" />
596 |           </folding>
597 |         </state>
598 |       </provider>
599 |     </entry>
600 |     <entry file="file://$PROJECT_DIR$/test.py">
601 |       <provider selected="true" editor-type-id="text-editor">
602 |         <state relative-caret-position="220">
603 |           <caret line="11" lean-forward="true" selection-start-line="11" selection-end-line="11" />
604 |           <folding>
605 |             <element signature="e#0#9#0" expanded="true" />
606 |           </folding>
607 |         </state>
608 |       </provider>
609 |     </entry>
610 |     <entry file="file://$PROJECT_DIR$/datasets.py">
611 |       <provider selected="true" editor-type-id="text-editor">
612 |         <state relative-caret-position="740">
613 |           <caret line="37" column="40" selection-start-line="37" selection-start-column="40" selection-end-line="37" selection-end-column="40" />
614 |           <folding>
615 |             <element signature="e#0#9#0" expanded="true" />
616 |           </folding>
617 |         </state>
618 |       </provider>
619 |     </entry>
620 |     <entry file="file://$PROJECT_DIR$/data_loaders.py">
621 |       <provider selected="true" editor-type-id="text-editor">
622 |         <state relative-caret-position="620">
623 |           <caret line="31" selection-start-line="31" selection-end-line="31" />
624 |         </state>
625 |       </provider>
626 |     </entry>
627 |     <entry file="file://$PROJECT_DIR$/utils.py">
628 |       <provider selected="true" editor-type-id="text-editor">
629 |         <state relative-caret-position="800">
630 |           <caret line="40" column="39" selection-start-line="40" selection-start-column="39" selection-end-line="40" selection-end-column="39" />
631 |           <folding>
632 |             <element signature="e#0#11#0" expanded="true" />
633 |           </folding>
634 |         </state>
635 |       </provider>
636 |     </entry>
637 |     <entry file="file://$PROJECT_DIR$/train.py">
638 |       <provider selected="true" editor-type-id="text-editor">
639 |         <state relative-caret-position="100">
640 |           <caret line="5" column="29" selection-start-line="5" selection-start-column="29" selection-end-line="5" selection-end-column="29" />
641 |         </state>
642 |       </provider>
643 |     </entry>
644 |     <entry file="file://$PROJECT_DIR$/models.py">
645 |       <provider selected="true" editor-type-id="text-editor">
646 |         <state relative-caret-position="2620">
647 |           <caret line="131" column="20" selection-start-line="131" selection-start-column="20" selection-end-line="131" selection-end-column="20" />
648 |           <folding>
649 |             <element signature="e#0#21#0" expanded="true" />
650 |           </folding>
651 |         </state>
652 |       </provider>
653 |     </entry>
654 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/torch/nn/modules/conv.py">
655 |       <provider selected="true" editor-type-id="text-editor">
656 |         <state relative-caret-position="6240">
657 |           <caret line="319" selection-start-line="319" selection-end-line="319" />
658 |         </state>
659 |       </provider>
660 |     </entry>
661 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/torch/nn/functional.py">
662 |       <provider selected="true" editor-type-id="text-editor">
663 |         <state relative-caret-position="32120">
664 |           <caret line="1622" selection-start-line="1622" selection-end-line="1622" />
665 |         </state>
666 |       </provider>
667 |     </entry>
668 |     <entry file="file://$PROJECT_DIR$/datasets.py">
669 |       <provider selected="true" editor-type-id="text-editor">
670 |         <state relative-caret-position="740">
671 |           <caret line="37" column="24" lean-forward="true" selection-start-line="37" selection-start-column="24" selection-end-line="37" selection-end-column="24" />
672 |           <folding>
673 |             <element signature="e#0#9#0" expanded="true" />
674 |           </folding>
675 |         </state>
676 |       </provider>
677 |     </entry>
678 |     <entry file="file://$PROJECT_DIR$/utils.py">
679 |       <provider selected="true" editor-type-id="text-editor">
680 |         <state relative-caret-position="140">
681 |           <caret line="7" column="7" lean-forward="true" selection-start-line="7" selection-start-column="7" selection-end-line="7" selection-end-column="7" />
682 |           <folding>
683 |             <element signature="e#0#11#0" expanded="true" />
684 |           </folding>
685 |         </state>
686 |       </provider>
687 |     </entry>
688 |     <entry file="file://$PROJECT_DIR$/losses.py">
689 |       <provider selected="true" editor-type-id="text-editor">
690 |         <state relative-caret-position="280">
691 |           <caret line="14" column="35" lean-forward="true" selection-start-line="14" selection-start-column="35" selection-end-line="14" selection-end-column="35" />
692 |           <folding>
693 |             <element signature="e#0#12#0" expanded="true" />
694 |           </folding>
695 |         </state>
696 |       </provider>
697 |     </entry>
698 |     <entry file="file://$PROJECT_DIR$/data_loaders.py" />
699 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/torch/nn/modules/conv.py">
700 |       <provider selected="true" editor-type-id="text-editor">
701 |         <state relative-caret-position="225">
702 |           <caret line="319" selection-start-line="319" selection-end-line="319" />
703 |         </state>
704 |       </provider>
705 |     </entry>
706 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/tqdm/_tqdm.py">
707 |       <provider selected="true" editor-type-id="text-editor">
708 |         <state relative-caret-position="225">
709 |           <caret line="1240" selection-start-line="1240" selection-end-line="1240" />
710 |         </state>
711 |       </provider>
712 |     </entry>
713 |     <entry file="file://$PROJECT_DIR$/models/baseline_0829_192301/0.png" />
714 |     <entry file="file://$PROJECT_DIR$/models/baseline_0829_192626/0.png" />
715 |     <entry file="file://$PROJECT_DIR$/models/baseline_0829_193049/0.png" />
716 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/torch/nn/functional.py">
717 |       <provider selected="true" editor-type-id="text-editor">
718 |         <state relative-caret-position="217">
719 |           <caret line="2076" selection-start-line="2076" selection-end-line="2076" />
720 |         </state>
721 |       </provider>
722 |     </entry>
723 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/matplotlib/axes/_subplots.py">
724 |       <provider selected="true" editor-type-id="text-editor">
725 |         <state relative-caret-position="225">
726 |           <caret line="64" selection-start-line="64" selection-end-line="64" />
727 |         </state>
728 |       </provider>
729 |     </entry>
730 |     <entry file="file:///anaconda3/envs/BEng/lib/python3.6/site-packages/matplotlib/image.py">
731 |       <provider selected="true" editor-type-id="text-editor">
732 |         <state relative-caret-position="225">
733 |           <caret line="637" selection-start-line="637" selection-end-line="637" />
734 |         </state>
735 |       </provider>
736 |     </entry>
737 |     <entry file="file://$PROJECT_DIR$/log.txt" />
738 |     <entry file="file://$PROJECT_DIR$/metrics.py">
739 |       <provider selected="true" editor-type-id="text-editor">
740 |         <state relative-caret-position="1960">
741 |           <caret line="98" column="17" selection-start-line="98" selection-start-column="17" selection-end-line="98" selection-end-column="17" />
742 |           <folding>
743 |             <element signature="e#0#18#0" expanded="true" />
744 |           </folding>
745 |         </state>
746 |       </provider>
747 |     </entry>
748 |     <entry file="file://$PROJECT_DIR$/models.py">
749 |       <provider selected="true" editor-type-id="text-editor">
750 |         <state relative-caret-position="2071">
751 |           <caret line="125" column="79" selection-start-line="125" selection-start-column="79" selection-end-line="125" selection-end-column="79" />
752 |           <folding>
753 |             <element signature="e#0#21#0" expanded="true" />
754 |           </folding>
755 |         </state>
756 |       </provider>
757 |     </entry>
758 |     <entry file="file://$PROJECT_DIR$/test.py">
759 |       <provider selected="true" editor-type-id="text-editor">
760 |         <state relative-caret-position="420">
761 |           <caret line="21" selection-start-line="21" selection-end-line="21" />
762 |           <folding>
763 |             <element signature="e#0#9#0" expanded="true" />
764 |           </folding>
765 |         </state>
766 |       </provider>
767 |     </entry>
768 |     <entry file="file://$PROJECT_DIR$/datasets.py">
769 |       <provider selected="true" editor-type-id="text-editor">
770 |         <state relative-caret-position="280">
771 |           <caret line="14" selection-start-line="14" selection-end-line="14" />
772 |           <folding>
773 |             <element signature="e#0#9#0" expanded="true" />
774 |           </folding>
775 |         </state>
776 |       </provider>
777 |     </entry>
778 |     <entry file="file://$PROJECT_DIR$/losses.py">
779 |       <provider selected="true" editor-type-id="text-editor">
780 |         <state relative-caret-position="513">
781 |           <caret line="66" column="19" selection-start-line="66" selection-start-column="19" selection-end-line="66" selection-end-column="19" />
782 |           <folding>
783 |             <element signature="e#0#12#0" expanded="true" />
784 |           </folding>
785 |         </state>
786 |       </provider>
787 |     </entry>
788 |     <entry file="file://$PROJECT_DIR$/train.py">
789 |       <provider selected="true" editor-type-id="text-editor">
790 |         <state relative-caret-position="553">
791 |           <caret line="112" column="24" selection-start-line="112" selection-start-column="24" selection-end-line="112" selection-end-column="24" />
792 |           <folding>
793 |             <element signature="e#0#9#0" expanded="true" />
794 |           </folding>
795 |         </state>
796 |       </provider>
797 |     </entry>
798 |     <entry file="file://$PROJECT_DIR$/utils.py">
799 |       <provider selected="true" editor-type-id="text-editor">
800 |         <state relative-caret-position="320">
801 |           <caret line="16" selection-start-line="16" selection-end-line="16" />
802 |           <folding>
803 |             <element signature="e#0#11#0" expanded="true" />
804 |           </folding>
805 |         </state>
806 |       </provider>
807 |     </entry>
808 |     <entry file="file://$PROJECT_DIR$/data_loaders.py">
809 |       <provider selected="true" editor-type-id="text-editor">
810 |         <state relative-caret-position="153">
811 |           <caret line="8" selection-start-line="8" selection-end-line="62" selection-end-column="40" />
812 |           <folding>
813 |             <element signature="e#0#18#0" expanded="true" />
814 |           </folding>
815 |         </state>
816 |       </provider>
817 |     </entry>
818 |   </component>
819 | </project>


--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
 1 | MIT License
 2 | 
 3 | Copyright (c) 2019 Hao
 4 | 
 5 | Permission is hereby granted, free of charge, to any person obtaining a copy
 6 | of this software and associated documentation files (the "Software"), to deal
 7 | in the Software without restriction, including without limitation the rights
 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 | 
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 | 
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # change_detection
2 |  Change Detection for high resolution statellite images
3 | 


--------------------------------------------------------------------------------
/data_loaders.py:
--------------------------------------------------------------------------------
 1 | import numpy as np
 2 | import torch
 3 | from torch.utils.data import DataLoader
 4 | from functools import partial
 5 | from datasets import RssraiDataset
 6 | from utils import calculate_bce_loss, get_pixels
 7 | 
 8 | 
 9 | def collate_fn(data, batch_size, img_size):
10 |     img, mask = data[0]
11 |     batch_x, batch_y = [], []
12 |     num_img = len(batch_x)
13 |     num_white_img = 0
14 | 
15 |     # randomly crop image and mask at white pixels
16 |     pixels = get_pixels(mask[0], img_size, img_size)
17 |     num_pix = len(pixels[0])
18 | 
19 |     while True:
20 |         if num_pix == 0:
21 |             s_x = np.random.randint(0, img.shape[1] - img_size + 1)
22 |             s_y = np.random.randint(0, img.shape[2] - img_size + 1)
23 |         else:
24 |             index = np.random.randint(num_pix)
25 |             s_x = pixels[0][index]
26 |             s_y = pixels[1][index]
27 |         y = mask[:, s_x:s_x + img_size, s_y:s_y + img_size]
28 | 
29 |         if len(np.where(y!= 0)[0]) > 0:
30 |             num_white_img += 1
31 | 
32 |         x = img[:, s_x:s_x + img_size, s_y:s_y + img_size]
33 | 
34 |         if num_white_img < 1:
35 |             continue
36 | 
37 |         batch_x.append(x)
38 |         batch_y.append(y)
39 |         num_img = len(batch_x)
40 | 
41 |         if num_img == batch_size:
42 |             break
43 | 
44 |     # for i in range(batch_size):
45 |     #     s_x = np.random.randint(0, img.shape[1] - img_size + 1)
46 |     #     s_y = np.random.randint(0, img.shape[2] - img_size + 1)
47 |     #     y = mask[:, s_x:s_x + img_size, s_y:s_y + img_size]
48 |     #     x = img[:, s_x:s_x + img_size, s_y:s_y + img_size]
49 |     #     batch_x.append(x)
50 |     #     batch_y.append(y)
51 | 
52 |     batch_x = np.array(batch_x)
53 |     batch_y = np.array(batch_y)
54 |     loss_weight = calculate_bce_loss(batch_y)
55 | 
56 |     if torch.cuda.is_available():
57 |         batch_x = torch.cuda.FloatTensor(batch_x)
58 |         batch_y = torch.cuda.FloatTensor(batch_y)
59 |     else:
60 |         batch_x = torch.FloatTensor(batch_x)
61 |         batch_y = torch.FloatTensor(batch_y)
62 | 
63 |     return batch_x, batch_y, loss_weight
64 | 
65 | 
66 | class RssraiDataLoader(DataLoader):
67 |     """
68 |     Retinal vessel segmentation data loader
69 |     """
70 |     def __init__(self,
71 |                  which_set='train',
72 |                  batch_size=16,
73 |                  img_size=256,
74 |                  shuffle=True
75 |                  ):
76 |         self.dataset = RssraiDataset(which_set=which_set)
77 |         self.batch_size = batch_size
78 |         self.img_size = img_size
79 |         self.shuffle = shuffle
80 | 
81 |         super(RssraiDataLoader, self).__init__(
82 |             dataset=self.dataset,
83 |             batch_size=1, # batch_size set to 1 as we use only 1 full images to extract many patches
84 |             shuffle=self.shuffle,
85 |             num_workers=0,
86 |             drop_last=self.shuffle,
87 |             collate_fn=partial(collate_fn, batch_size=self.batch_size, img_size=self.img_size))
88 | 
89 | 
90 | if __name__ == '__main__':
91 |     data_loader = RssraiDataLoader(which_set='train', batch_size=16, img_size=256, shuffle=True)
92 | 
93 |     for i, (input, mask, loss_weight) in enumerate(data_loader):
94 |         print('{}th batch: input shape {}, mask shape {}, loss_weight{}'.format(i, input.shape, mask.shape, loss_weight))
95 | 
96 | 


--------------------------------------------------------------------------------
/datasets.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import numpy as np
 3 | from tifffile import imread
 4 | from torch.utils.data import Dataset
 5 | from utils import get_files
 6 | 
 7 | 
 8 | class RssraiDataset(Dataset):
 9 |     """
10 |     rssrai2019 change detection dataset
11 |     """
12 |     def __init__(self, which_set):
13 |         self.data_dir = os.path.join(os.path.dirname(__file__), 'data/rssrai2019_change_detection', which_set)
14 |         assert os.path.exists(self.data_dir), "cannot find data folder: {}".format(self.data_dir)
15 | 
16 |         self.img1_ids = get_files(os.path.join(self.data_dir, 'img_2017/'))
17 |         self.img2_ids = get_files(os.path.join(self.data_dir, 'img_2018/'))
18 |         self.mask_ids = get_files(os.path.join(self.data_dir, 'mask/'))
19 | 
20 |     def __len__(self):
21 |         return len(self.img1_ids)
22 | 
23 |     def __getitem__(self, idx):
24 |         img1_id = self.img1_ids[idx]
25 |         img2_id = self.img2_ids[idx]
26 |         mask_id = self.mask_ids[idx]
27 |         img1 = imread(img1_id).astype('uint8')
28 |         img2 = imread(img2_id).astype('uint8')
29 |         mask = imread(mask_id).astype('uint8')
30 | 
31 |         img1 = img1.astype('float32') / 255.
32 |         img2 = img2.astype('float32') / 255.
33 |         mask = mask.astype('float32') / 255.
34 | 
35 |         img = np.concatenate([img1, img2], axis=-1)
36 |         img = img.transpose((2, 0, 1))
37 |         mask = mask[:, :, np.newaxis]
38 |         mask = mask.transpose((2, 0, 1))
39 |         return img, mask
40 | 
41 | 
42 | if __name__ == '__main__':
43 |     from tqdm import tqdm
44 |     import matplotlib.pyplot as plt
45 |     import cv2
46 | 
47 |     train_dataset = RssraiDataset(which_set='train')
48 |     print('length of the dataset: {}'.format(len(train_dataset)))
49 | 
50 |     for i, (input, mask) in tqdm(enumerate(train_dataset), total=len(train_dataset)):
51 |         print('input image shape:{}'.format(input.shape))
52 |         print('mask shape:{}'.format(mask.shape))
53 |         input = input.transpose(1, 2, 0)
54 | 
55 |         plt.imshow((input[:, :, :3] * 255).astype('uint8'))
56 |         plt.show()
57 | 
58 |         cv2.imshow('image', input[:, :, :3] * 255)
59 |         cv2.waitKey(0)
60 |         # cv2.destroyAllWindows()
61 | 
62 |         break
63 | 
64 | 


--------------------------------------------------------------------------------
/losses.py:
--------------------------------------------------------------------------------
 1 | import torch
 2 | import torch.nn as nn
 3 | import torch.nn.functional as F
 4 | 
 5 | 
 6 | def weighted_binary_cross_entropy(sigmoid_x, targets, pos_weight, weight=None, size_average=True, reduce=True):
 7 |     """
 8 |     Args:
 9 |         sigmoid_x: predicted probability of size [N,C], N sample and C Class. Eg. Must be in range of [0,1], i.e. Output from Sigmoid.
10 |         targets: true value, one-hot-like vector of size [N,C]
11 |         pos_weight: Weight for postive sample
12 |     """
13 |     if not (targets.size() == sigmoid_x.size()):
14 |         raise ValueError("Target size ({}) must be the same as input size ({})".format(targets.size(), sigmoid_x.size()))
15 | 
16 |     sigmoid_x = sigmoid_x.clamp(min=1e-8, max=1-1e-8)
17 |     loss = -pos_weight * targets * sigmoid_x.log() - (1-targets)*(1-sigmoid_x).log()
18 | 
19 |     if weight is not None:
20 |         loss = loss * weight
21 | 
22 |     if not reduce:
23 |         return loss
24 |     elif size_average:
25 |         return loss.mean()
26 |     else:
27 |         return loss.sum()
28 | 
29 | 
30 | def dice_loss(pred, target, smooth=1e-8):
31 |     iflat = pred.view(-1)
32 |     tflat = target.view(-1)
33 |     intersection = (iflat * tflat).sum()
34 | 
35 |     return 1 - ((2. * intersection + smooth) /
36 |                 (iflat.sum() + tflat.sum() + smooth))
37 | 
38 | 
39 | class WBCEDiceLoss(nn.Module):
40 |     def __init__(self, alpha=0.5):
41 |         super(WBCEDiceLoss, self).__init__()
42 |         self.alpha = alpha
43 | 
44 |     def forward(self, pred, target, weight):
45 |         pred = F.sigmoid(pred)
46 |         dice = dice_loss(pred, target)
47 |         bce = weighted_binary_cross_entropy(pred, target, weight)
48 | 
49 |         # bce = F.binary_cross_entropy_with_logits(pred, target)
50 |         # pred = F.sigmoid(pred)
51 | 
52 |         loss = bce + dice * self.alpha
53 |         return loss
54 | 
55 | 
56 | class BCEDiceLoss(nn.Module):
57 |     def __init__(self, alpha=0.5):
58 |         super(BCEDiceLoss, self).__init__()
59 |         self.alpha = alpha
60 | 
61 |     def forward(self, pred, target, weight):
62 |         bce = F.binary_cross_entropy_with_logits(pred, target)
63 |         pred = F.sigmoid(pred)
64 |         dice = dice_loss(pred, target)
65 | 
66 |         loss = bce + dice * self.alpha
67 |         return loss
68 |     


--------------------------------------------------------------------------------
/metrics.py:
--------------------------------------------------------------------------------
  1 | import numpy as np
  2 | 
  3 | import torch
  4 | import torch.nn.functional as F
  5 | 
  6 | 
  7 | def mean_iou(y_true_in, y_pred_in, print_table=False):
  8 |     if True: #not np.sum(y_true_in.flatten()) == 0:
  9 |         labels = y_true_in
 10 |         y_pred = y_pred_in
 11 | 
 12 |         true_objects = 2
 13 |         pred_objects = 2
 14 | 
 15 |         intersection = np.histogram2d(labels.flatten(), y_pred.flatten(), bins=(true_objects, pred_objects))[0]
 16 | 
 17 |         # Compute areas (needed for finding the union between all objects)
 18 |         area_true = np.histogram(labels, bins = true_objects)[0]
 19 |         area_pred = np.histogram(y_pred, bins = pred_objects)[0]
 20 |         area_true = np.expand_dims(area_true, -1)
 21 |         area_pred = np.expand_dims(area_pred, 0)
 22 | 
 23 |         # Compute union
 24 |         union = area_true + area_pred - intersection
 25 | 
 26 |         # Exclude background from the analysis
 27 |         intersection = intersection[1:,1:]
 28 |         union = union[1:,1:]
 29 |         union[union == 0] = 1e-9
 30 | 
 31 |         # Compute the intersection over union
 32 |         iou = intersection / union
 33 | 
 34 |         # Precision helper function
 35 |         def precision_at(threshold, iou):
 36 |             matches = iou > threshold
 37 |             true_positives = np.sum(matches, axis=1) == 1   # Correct objects
 38 |             false_positives = np.sum(matches, axis=0) == 0  # Missed objects
 39 |             false_negatives = np.sum(matches, axis=1) == 0  # Extra objects
 40 |             tp, fp, fn = np.sum(true_positives), np.sum(false_positives), np.sum(false_negatives)
 41 |             return tp, fp, fn
 42 | 
 43 |         # Loop over IoU thresholds
 44 |         prec = []
 45 |         if print_table:
 46 |             print("Thresh\tTP\tFP\tFN\tPrec.")
 47 |         for t in np.arange(0.5, 1.0, 0.05):
 48 |             tp, fp, fn = precision_at(t, iou)
 49 |             if (tp + fp + fn) > 0:
 50 |                 p = tp / (tp + fp + fn)
 51 |             else:
 52 |                 p = 0
 53 |             if print_table:
 54 |                 print("{:1.3f}\t{}\t{}\t{}\t{:1.3f}".format(t, tp, fp, fn, p))
 55 |             prec.append(p)
 56 | 
 57 |         if print_table:
 58 |             print("AP\t-\t-\t-\t{:1.3f}".format(np.mean(prec)))
 59 |         return np.mean(prec)
 60 | 
 61 |     else:
 62 |         if np.sum(y_pred_in.flatten()) == 0:
 63 |             return 1
 64 |         else:
 65 |             return 0
 66 | 
 67 | 
 68 | def batch_iou(output, target):
 69 |     output = torch.sigmoid(output).data.cpu().numpy() > 0.5
 70 |     target = (target.data.cpu().numpy() > 0.5).astype('int')
 71 |     output = output[:,0,:,:]
 72 |     target = target[:,0,:,:]
 73 | 
 74 |     ious = []
 75 |     for i in range(output.shape[0]):
 76 |         ious.append(mean_iou(output[i], target[i]))
 77 | 
 78 |     return np.mean(ious)
 79 | 
 80 | 
 81 | def mean_iou(output, target):
 82 |     smooth = 1e-5
 83 | 
 84 |     output = torch.sigmoid(output).data.cpu().numpy()
 85 |     target = target.data.cpu().numpy()
 86 |     ious = []
 87 |     for t in np.arange(0.5, 1.0, 0.05):
 88 |         output_ = output > t
 89 |         target_ = target > t
 90 |         intersection = (output_ & target_).sum()
 91 |         union = (output_ | target_).sum()
 92 |         iou = (intersection + smooth) / (union + smooth)
 93 |         ious.append(iou)
 94 | 
 95 |     return np.mean(ious)
 96 | 
 97 | 
 98 | def iou_score(output, target):
 99 |     smooth = 1e-5
100 | 
101 |     if torch.is_tensor(output):
102 |         output = torch.sigmoid(output).data.cpu().numpy()
103 |     if torch.is_tensor(target):
104 |         target = target.data.cpu().numpy()
105 |     output_ = output > 0.5
106 |     target_ = target > 0.5
107 |     intersection = (output_ & target_).sum()
108 |     union = (output_ | target_).sum()
109 | 
110 |     return (intersection + smooth) / (union + smooth)
111 | 
112 | 
113 | def dice_coef(output, target):
114 |     smooth = 1e-5
115 | 
116 |     output = torch.sigmoid(output).view(-1).data.cpu().numpy()
117 |     target = target.view(-1).data.cpu().numpy()
118 |     intersection = (output * target).sum()
119 | 
120 |     return (2. * intersection + smooth) / \
121 |         (output.sum() + target.sum() + smooth)
122 | 
123 | 
124 | def accuracy(output, target):
125 |     output = torch.sigmoid(output).view(-1).data.cpu().numpy()
126 |     output = (np.round(output)).astype('int')
127 |     target = target.view(-1).data.cpu().numpy()
128 |     target = (np.round(target)).astype('int')
129 |     (output == target).sum()
130 | 
131 |     return (output == target).sum() / len(output)
132 | 


--------------------------------------------------------------------------------
/models.py:
--------------------------------------------------------------------------------
  1 | import torch.nn as nn
  2 | import torch
  3 | 
  4 | 
  5 | class ConvBlock(nn.Module):
  6 |     """
  7 |     Convolution Block
  8 |     """
  9 |     def __init__(self, in_ch, out_ch, kernel_size=3, stride=1, padding=1, dilation=1):
 10 |         super(ConvBlock, self).__init__()
 11 |         block = [nn.Conv2d(in_channels=in_ch, out_channels=out_ch, kernel_size=kernel_size, stride=stride, padding=padding, dilation=dilation, bias=True)]
 12 |         block += [nn.BatchNorm2d(out_channels=out_ch)]
 13 |         block += [nn.ReLU(inplace=True)]
 14 |         self.block = nn.Sequential(*block)
 15 | 
 16 |     def forward(self, x):
 17 |         return self.block(x)
 18 | 
 19 | 
 20 | class UpConvBlock(nn.Module):
 21 |     """
 22 |     Upsampling Convolution Block
 23 |     """
 24 |     def __init__(self, in_ch, out_ch, kernel_size=3, stride=1, padding=1, dilation=1):
 25 |         super(UpConvBlock, self).__init__()
 26 |         block = [nn.Upsample(scale_factor=2)]
 27 |         block += [nn.Conv2d(in_channels=in_ch, out_channels=out_ch, kernel_size=kernel_size, stride=stride, padding=padding, dilation=dilation, bias=True)]
 28 |         block += [nn.BatchNorm2d(out_channels=out_ch)]
 29 |         block += [nn.ReLU(inplace=True)]
 30 |         self.block = nn.Sequential(*block)
 31 | 
 32 |     def forward(self, x):
 33 |         return self.block(x)
 34 | 
 35 | 
 36 | class ResBlock(nn.Module):
 37 |     """
 38 |     Residual Block
 39 |     """
 40 |     def __init__(self, in_ch, out_ch, stride=1):
 41 |         super(ResBlock, self).__init__()
 42 |         self.relu = nn.ReLU(inplace=True)
 43 |         self.conv1 = nn.Conv2d(in_channels=in_ch, out_channels=out_ch, kernel_size=3, stride=stride, padding=1, dilation=1, bias=True)
 44 |         self.bn1 = nn.BatchNorm2d(out_ch)
 45 |         self.conv2 = nn.Conv2d(in_channels=out_ch, out_channels=out_ch, kernel_size=3, stride=stride, padding=1, dilation=1, bias=True)
 46 |         self.bn2 = nn.BatchNorm2d(out_ch)
 47 | 
 48 |     def forward(self, x):
 49 |         identity = self.conv1(x)
 50 |         out = self.bn1(identity)
 51 |         out = self.relu(out)
 52 | 
 53 |         out = self.conv2(out)
 54 |         out = self.bn2(out)
 55 | 
 56 |         out += identity
 57 |         out = self.relu(out)
 58 |         return out
 59 | 
 60 | 
 61 | class NestedUNet(nn.Module):
 62 |     """
 63 |     Implementation of nested Unet (Unet++)
 64 |     """
 65 |     
 66 |     def __init__(self, in_ch=3, out_ch=1, n=32):
 67 |         super(NestedUNet, self).__init__()
 68 | 
 69 |         filters = [n, n * 2, n * 4, n * 8, n * 16]
 70 | 
 71 |         self.conv0_0 = ResBlock(in_ch, filters[0])
 72 |         self.conv1_0 = ResBlock(filters[0], filters[1])
 73 |         self.conv2_0 = ResBlock(filters[1], filters[2])
 74 |         self.conv3_0 = ResBlock(filters[2], filters[3])
 75 |         self.conv4_0 = ResBlock(filters[3], filters[4])
 76 | 
 77 |         self.conv0_1 = ResBlock(filters[0] + filters[1], filters[0])
 78 |         self.conv1_1 = ResBlock(filters[1] + filters[2], filters[1])
 79 |         self.conv2_1 = ResBlock(filters[2] + filters[3], filters[2])
 80 |         self.conv3_1 = ResBlock(filters[3] + filters[4], filters[3])
 81 | 
 82 |         self.conv0_2 = ResBlock(filters[0] * 2 + filters[1], filters[0])
 83 |         self.conv1_2 = ResBlock(filters[1] * 2 + filters[2], filters[1])
 84 |         self.conv2_2 = ResBlock(filters[2] * 2 + filters[3], filters[2])
 85 | 
 86 |         self.conv0_3 = ResBlock(filters[0] * 3 + filters[1], filters[0])
 87 |         self.conv1_3 = ResBlock(filters[1] * 3 + filters[2], filters[1])
 88 | 
 89 |         self.conv0_4 = ResBlock(filters[0] * 4 + filters[1], filters[0])
 90 | 
 91 |         self.pool = nn.MaxPool2d(kernel_size=2, stride=2)
 92 |         self.up = nn.Upsample(scale_factor=2, mode='bilinear', align_corners=True)
 93 | 
 94 |         self.final1 = nn.Conv2d(filters[0], out_ch, kernel_size=1)
 95 |         self.final2 = nn.Conv2d(filters[0], out_ch, kernel_size=1)
 96 |         self.final3 = nn.Conv2d(filters[0], out_ch, kernel_size=1)
 97 |         self.final4 = nn.Conv2d(filters[0], out_ch, kernel_size=1)
 98 |         self.final5 = nn.Conv2d(4, out_ch, kernel_size=1)
 99 | 
100 |     def forward(self, x):
101 | 
102 |         x0_0 = self.conv0_0(x)
103 | 
104 |         x1_0 = self.conv1_0(self.pool(x0_0))
105 |         x0_1 = self.conv0_1(torch.cat((x0_0, self.up(x1_0)), 1))
106 | 
107 |         x2_0 = self.conv2_0(self.pool(x1_0))
108 |         x1_1 = self.conv1_1(torch.cat((x1_0, self.up(x2_0)), 1))
109 |         x0_2 = self.conv0_2(torch.cat((x0_0, x0_1, self.up(x1_1)), 1))
110 | 
111 |         x3_0 = self.conv3_0(self.pool(x2_0))
112 |         x2_1 = self.conv2_1(torch.cat((x2_0, self.up(x3_0)), 1))
113 |         x1_2 = self.conv1_2(torch.cat((x1_0, x1_1, self.up(x2_1)), 1))
114 |         x0_3 = self.conv0_3(torch.cat((x0_0, x0_1, x0_2, self.up(x1_2)), 1))
115 | 
116 |         x4_0 = self.conv4_0(self.pool(x3_0))
117 |         x3_1 = self.conv3_1(torch.cat((x3_0, self.up(x4_0)), 1))
118 |         x2_2 = self.conv2_2(torch.cat((x2_0, x2_1, self.up(x3_1)), 1))
119 |         x1_3 = self.conv1_3(torch.cat((x1_0, x1_1, x1_2, self.up(x2_2)), 1))
120 |         x0_4 = self.conv0_4(torch.cat((x0_0, x0_1, x0_2, x0_3, self.up(x1_3)), 1))
121 | 
122 |         output1 = self.final1(x0_1)
123 |         output2 = self.final2(x0_2)
124 |         output3 = self.final3(x0_3)
125 |         output4 = self.final4(x0_4)
126 |         output5 = self.final5(torch.cat((output1, output2, output3, output4), 1))
127 | 
128 |         return [output1, output2, output3, output4, output5]
129 | 
130 | 
131 | if __name__ == '__main__':
132 |     a = torch.ones((16, 6, 256, 256))
133 |     Unet = NestedUNet(in_ch=6, out_ch=1)
134 |     output = Unet(a)


--------------------------------------------------------------------------------
/test.py:
--------------------------------------------------------------------------------
  1 | import os
  2 | import argparse
  3 | from glob import glob
  4 | import warnings
  5 | 
  6 | import numpy as np
  7 | from tqdm import tqdm
  8 | 
  9 | from skimage.io import imread, imsave
 10 | 
 11 | import torch
 12 | 
 13 | import models
 14 | from metrics import dice_coef, batch_iou, mean_iou, iou_score
 15 | import losses
 16 | from utils import count_params
 17 | from data_loaders import RssraiDataLoader
 18 | 
 19 | 
 20 | def parse_args():
 21 |     parser = argparse.ArgumentParser()
 22 | 
 23 |     parser.add_argument('--name', default=None,
 24 |                         help='model name')
 25 | 
 26 |     args = parser.parse_args()
 27 | 
 28 |     return args
 29 | 
 30 | 
 31 | def main():
 32 |     val_args = parse_args()
 33 | 
 34 |     args = joblib.load('models/%s/args.pkl' %val_args.name)
 35 | 
 36 |     if not os.path.exists('output/%s' %args.name):
 37 |         os.makedirs('output/%s' %args.name)
 38 | 
 39 |     print('Config -----')
 40 |     for arg in vars(args):
 41 |         print('%s: %s' %(arg, getattr(args, arg)))
 42 |     print('------------')
 43 | 
 44 |     joblib.dump(args, 'models/%s/args.pkl' %args.name)
 45 | 
 46 |     # create model
 47 |     print("=> creating model %s" %args.arch)
 48 |     model = models.__dict__[args.arch](args.in_ch, args.out_ch, args.num_filters)
 49 | 
 50 |     if torch.cuda.is_available():
 51 |         model = model.cuda()
 52 | 
 53 |     model.load_state_dict(torch.load('models/%s/model.pth' %args.name))
 54 |     model.eval()
 55 | 
 56 |     val_loader = RssraiDataLoader(
 57 |         which_set='test',
 58 |         batch_size=args.batch_size,
 59 |         img_size=args.img_size,
 60 |         shuffle=False
 61 |     )
 62 | 
 63 |     with warnings.catch_warnings():
 64 |         warnings.simplefilter('ignore')
 65 | 
 66 |         with torch.no_grad():
 67 |             for i, (input, target) in tqdm(enumerate(val_loader), total=len(val_loader)):
 68 |                 # compute output
 69 |                 output = model(input)[-1]
 70 |                 output = torch.sigmoid(output).data.cpu().numpy()
 71 |                 img_paths = val_img_paths[args.batch_size*i:args.batch_size*(i+1)]
 72 | 
 73 |                 for i in range(output.shape[0]):
 74 |                     imsave('output/%s/'%args.name+os.path.basename(img_paths[i]), (output[i,0,:,:]*255).astype('uint8'))
 75 | 
 76 |         torch.cuda.empty_cache()
 77 | 
 78 |     # IoU
 79 |     ious = []
 80 |     for i in tqdm(range(len(val_mask_paths))):
 81 |         mask = imread(val_mask_paths[i])
 82 |         pb = imread('output/%s/'%args.name+os.path.basename(val_mask_paths[i]))
 83 | 
 84 |         mask = mask.astype('float32') / 255
 85 |         pb = pb.astype('float32') / 255
 86 | 
 87 | 
 88 |         '''
 89 |         plt.figure()
 90 |         plt.subplot(121)
 91 |         plt.imshow(mask)
 92 |         plt.subplot(122)
 93 |         plt.imshow(pb)
 94 |         plt.show()
 95 |         '''
 96 | 
 97 |         iou = iou_score(pb, mask)
 98 |         ious.append(iou)
 99 |     print('IoU: %.4f' %np.mean(ious))
100 | 
101 | 
102 | if __name__ == '__main__':
103 |     main()


--------------------------------------------------------------------------------
/train.py:
--------------------------------------------------------------------------------
  1 | import os
  2 | import argparse
  3 | import pandas as pd
  4 | from tqdm import tqdm
  5 | from collections import OrderedDict
  6 | from datetime import datetime
  7 | from sklearn.externals import joblib
  8 | 
  9 | import torch
 10 | import torch.nn as nn
 11 | import torch.optim as optim
 12 | import torch.backends.cudnn as cudnn
 13 | from torch.optim.lr_scheduler import CosineAnnealingLR
 14 | 
 15 | import models
 16 | import losses
 17 | from data_loaders import RssraiDataLoader
 18 | from metrics import iou_score
 19 | from utils import count_params, save_example
 20 | 
 21 | arch_names = list(models.__dict__.keys())
 22 | loss_names = list(losses.__dict__.keys())
 23 | loss_names.append('BCEWithLogitsLoss')
 24 | 
 25 | 
 26 | def parse_args():
 27 |     parser = argparse.ArgumentParser()
 28 | 
 29 |     parser.add_argument('--name', default='baseline',
 30 |                         help='model name: (default: arch+timestamp)')
 31 |     parser.add_argument('--img_size', default=256, type=int, help='size of training image patches')
 32 |     parser.add_argument('--arch', '-a', metavar='ARCH', default='NestedUNet',
 33 |                         choices=arch_names,
 34 |                         help='model architecture: ' +
 35 |                             ' | '.join(arch_names) +
 36 |                             ' (default: NestedUNet)')
 37 |     parser.add_argument('--num_filters', default=32, type=int,
 38 |                         help='number of starting filters in CNN.')
 39 |     parser.add_argument('--in_ch', default=8, type=int,
 40 |                         help='input channels')
 41 |     parser.add_argument('--out_ch', default=1, type=int,
 42 |                         help='output channels')
 43 |     parser.add_argument('--loss', default='BCEDiceLoss',
 44 |                         choices=loss_names,
 45 |                         help='loss: ' +
 46 |                             ' | '.join(loss_names) +
 47 |                             ' (default: BCEDiceLoss)')
 48 |     parser.add_argument('--epochs', default=1000, type=int, metavar='N',
 49 |                         help='number of total epochs to run')
 50 |     parser.add_argument('--early-stop', default=20, type=int,
 51 |                         metavar='N', help='early stopping (default: 20)')
 52 |     parser.add_argument('-b', '--batch_size', default=16, type=int,
 53 |                         metavar='N', help='mini-batch size (default: 16)')
 54 |     parser.add_argument('--optimizer', default='Adam',
 55 |                         choices=['Adam', 'SGD'],
 56 |                         help='loss: ' +
 57 |                             ' | '.join(['Adam', 'SGD']) +
 58 |                             ' (default: Adam)')
 59 |     parser.add_argument('--lr', '--learning-rate', default=2e-4, type=float,
 60 |                         metavar='LR', help='initial learning rate')
 61 |     parser.add_argument('--momentum', default=0.9, type=float,
 62 |                         help='momentum')
 63 |     parser.add_argument('--weight-decay', default=1e-6, type=float,
 64 |                         help='weight decay')
 65 |     args = parser.parse_args()
 66 | 
 67 |     return args
 68 | 
 69 | 
 70 | class AverageMeter(object):
 71 |     """Computes and stores the average and current value"""
 72 |     def __init__(self):
 73 |         self.reset()
 74 | 
 75 |     def reset(self):
 76 |         self.val = 0
 77 |         self.avg = 0
 78 |         self.sum = 0
 79 |         self.count = 0
 80 | 
 81 |     def update(self, val, n=1):
 82 |         self.val = val
 83 |         self.sum += val * n
 84 |         self.count += n
 85 |         self.avg = self.sum / self.count
 86 | 
 87 | 
 88 | def train(train_loader, model, criterion, optimizer):
 89 |     losses = AverageMeter()
 90 |     ious = AverageMeter()
 91 | 
 92 |     model.train()
 93 | 
 94 |     pbar = tqdm(enumerate(train_loader), total=len(train_loader))
 95 | 
 96 |     for i, (input, target, loss_weight) in pbar:
 97 |         # compute output
 98 |         outputs = model(input)
 99 |         loss = 0
100 |         for output in outputs:
101 |             loss += criterion(output, target, loss_weight)
102 |         loss /= len(outputs)
103 |         iou = iou_score(outputs[-1], target)
104 | 
105 |         # update log and progress bar
106 |         losses.update(loss.item(), input.size(0))
107 |         ious.update(iou, input.size(0))
108 |         pbar.set_postfix({'loss': loss.item(), 'iou': iou})
109 | 
110 |         # compute gradient and do optimizing step
111 |         optimizer.zero_grad()
112 |         loss.backward()
113 |         optimizer.step()
114 | 
115 |     log = OrderedDict([
116 |         ('loss', losses.avg),
117 |         ('iou', ious.avg),
118 |     ])
119 | 
120 |     return log
121 | 
122 | 
123 | def validate(val_loader, model, criterion):
124 |     losses = AverageMeter()
125 |     ious = AverageMeter()
126 | 
127 |     # switch to evaluate mode
128 |     model.eval()
129 | 
130 |     with torch.no_grad():
131 |         for i, (input, target, loss_weight) in enumerate(val_loader):
132 |             # compute output
133 |             outputs = model(input)
134 |             loss = 0
135 |             for output in outputs:
136 |                 loss += criterion(output, target, loss_weight)
137 |             loss /= len(outputs)
138 |             iou = iou_score(outputs[-1], target)
139 | 
140 |             losses.update(loss.item(), input.size(0))
141 |             ious.update(iou, input.size(0))
142 | 
143 |     log = OrderedDict([
144 |         ('loss', losses.avg),
145 |         ('iou', ious.avg),
146 |     ])
147 | 
148 |     return log
149 | 
150 | 
151 | def main():
152 |     args = parse_args()
153 |     timestamp = datetime.now().strftime('%m%d_%H%M%S')
154 |     model_dir = 'models/{}'.format(args.name + '_' + timestamp)
155 | 
156 |     # check if model directory if exists
157 |     if not os.path.exists(model_dir):
158 |         os.makedirs(model_dir)
159 | 
160 |     # print configuration and save it
161 |     print('Config -----')
162 |     for arg in vars(args):
163 |         print('{0}: {1}'.format(arg, getattr(args, arg)))
164 |     print('------------')
165 | 
166 |     with open('{}/args.txt'.format(model_dir), 'w') as f:
167 |         for arg in vars(args):
168 |             print('%s: %s' %(arg, getattr(args, arg)), file=f)
169 | 
170 |     joblib.dump(args, '{}/args.pkl'.format(model_dir))
171 | 
172 |     # define loss function (criterion)
173 |     if args.loss == 'BCEWithLogitsLoss':
174 |         criterion = nn.BCEWithLogitsLoss().cuda()
175 |     else:
176 |         criterion = losses.__dict__[args.loss]().cuda()
177 | 
178 |     # create model
179 |     print("=> creating model {}".format(args.arch))
180 |     model = models.__dict__[args.arch](args.in_ch, args.out_ch, args.num_filters)
181 | 
182 |     if torch.cuda.is_available():
183 |         cudnn.benchmark = True
184 |         model = model.cuda()
185 | 
186 |     print(count_params(model))
187 | 
188 |     if args.optimizer == 'Adam':
189 |         optimizer = optim.Adam(filter(lambda p: p.requires_grad, model.parameters()), lr=args.lr)
190 |     elif args.optimizer == 'SGD':
191 |         optimizer = optim.SGD(filter(lambda p: p.requires_grad, model.parameters()), lr=args.lr,
192 |             momentum=args.momentum, weight_decay=args.weight_decay)
193 |     else:
194 |         raise ValueError('optimizer not specified')
195 | 
196 |     scheduler = CosineAnnealingLR(optimizer, T_max=100, eta_min=1e-8)
197 | 
198 |     train_loader = RssraiDataLoader(which_set='train', batch_size=args.batch_size, img_size=args.img_size, shuffle=True)
199 |     val_loader = RssraiDataLoader(which_set='val', batch_size=1, img_size=960, shuffle=True)
200 | 
201 |     log = pd.DataFrame(index=[], columns=[
202 |         'epoch', 'lr', 'loss', 'iou', 'val_loss', 'val_iou'
203 |     ])
204 | 
205 |     best_loss = 0
206 |     trigger = 0
207 | 
208 |     for epoch in range(args.epochs):
209 |         print('Epoch [{0:d}/{1:d}]'.format(epoch, args.epochs))
210 | 
211 |         # train for one epoch
212 |         train_log = train(train_loader=train_loader, model=model, criterion=criterion, optimizer=optimizer)
213 |         # evaluate on validation set
214 |         val_log = validate(val_loader=val_loader, model=model, criterion=criterion)
215 |         # update learning rate
216 |         scheduler.step()
217 | 
218 |         print('loss {0:.4f} - iou {1:.4f} - val_loss {2:.4f} - val_iou {3:.4f}'.format(train_log['loss'], train_log['iou'], val_log['loss'], val_log['iou']))
219 | 
220 |         tmp = pd.Series([
221 |             epoch,
222 |             args.lr,
223 |             train_log['loss'],
224 |             train_log['iou'],
225 |             val_log['loss'],
226 |             val_log['iou'],
227 |         ], index=['epoch', 'lr', 'loss', 'iou', 'val_loss', 'val_iou'])
228 | 
229 |         save_example(model_dir, epoch, model, val_loader)
230 |         log = log.append(tmp, ignore_index=True)
231 |         log.to_csv(os.path.join(model_dir, 'log.csv'), index=False)
232 | 
233 |         trigger += 1
234 | 
235 |         if epoch == 0:
236 |             best_loss = val_log['loss']
237 | 
238 |         if val_log['loss'] <= best_loss:
239 |             torch.save(model.state_dict(), os.path.join(model_dir, 'model_best.pth'))
240 |             best_loss = val_log['loss']
241 |             print("=> saved best model")
242 |             trigger = 0
243 | 
244 |         # early stopping
245 |         if not args.early_stop is None:
246 |             if trigger >= args.early_stop:
247 |                 print("=> early stopping")
248 |                 break
249 |         torch.save(model.state_dict(), os.path.join(model_dir, 'model_final.pth'))
250 | 
251 |         torch.cuda.empty_cache()
252 | 
253 | 
254 | if __name__ == '__main__':
255 |     main()
256 | 
257 | 
258 | 
259 | 
260 | 


--------------------------------------------------------------------------------
/utils.py:
--------------------------------------------------------------------------------
  1 | import glob
  2 | import numpy as np
  3 | import torch
  4 | import matplotlib.pyplot as plt
  5 | 
  6 | 
  7 | def calculate_bce_loss(masks, epsilon=1e-8):
  8 |     y_minus = 0
  9 |     y_plus = 0
 10 |     single_total = masks.shape[-1] * masks.shape[-2]
 11 |     for m in masks:
 12 |         black = len(np.where(m[0]==0)[0])
 13 |         white = single_total - black
 14 |         y_minus += black
 15 |         y_plus += white
 16 | 
 17 |     return y_minus / (y_plus + epsilon)
 18 | 
 19 | 
 20 | def get_pixels(image, boarder_height, boarder_width):
 21 |     """
 22 |     Get the white pixels from an 2D image, with boarder removed defined by boarder height and boarder width
 23 |     :param image: 2D image array
 24 |     :param boarder_height: boarder height to be removed
 25 |     :param boarder_width: boarder width to be removed
 26 |     :return: points
 27 |     """
 28 |     image_height = image.shape[0]
 29 |     image_width = image.shape[1]
 30 |     masked_image = image.copy()
 31 |     mask = np.zeros_like(image)
 32 |     mask[:image_height-boarder_height, :image_width-boarder_width] = 1
 33 |     masked_image[mask == 0] = 0
 34 |     points = np.where(masked_image != 0)
 35 |     return points
 36 | 
 37 | 
 38 | def get_files(directory, format='tif'):
 39 |     """
 40 |     To get a list of file names in one directory, especially images
 41 |     :param directory: a path to the directory of the image files
 42 |     :return: a list of all the file names in that directory
 43 |     """
 44 |     if format is 'png':
 45 |         file_list = glob.glob(directory + "*.png")
 46 |     elif format is 'tif':
 47 |         file_list = glob.glob(directory + "*.tif")
 48 |     else:
 49 |         raise ValueError("dataset do not support")
 50 | 
 51 |     file_list.sort(key=lambda f: int(''.join(filter(str.isdigit, f))))
 52 |     return file_list
 53 | 
 54 | 
 55 | def count_params(model):
 56 |     return sum(p.numel() for p in model.parameters() if p.requires_grad)
 57 | 
 58 | 
 59 | def save_example(folder, epoch, model, data_loader):
 60 |     """
 61 |     save prediction examples during training
 62 |     """
 63 |     if epoch == 0 or epoch % 5 == 0:
 64 |         model.eval()
 65 |         with torch.no_grad():
 66 |             inputs, targets, _ = next(iter(data_loader))
 67 |             # if torch.cuda.is_available():
 68 |             #     inputs = inputs.cuda()
 69 | 
 70 |             preds = model(inputs)
 71 |             preds = preds[-1]
 72 |             preds = torch.sigmoid(preds)
 73 | 
 74 |             if torch.cuda.is_available():
 75 |                 preds = preds.cpu()
 76 |                 inputs = inputs.cpu()
 77 |                 targets = targets.cpu()
 78 | 
 79 |             inputs, targets, preds = inputs.numpy(), targets.numpy(), preds.numpy()
 80 |             inputs, targets, preds = (inputs * 255).astype('uint8'), (targets * 255).astype('uint8'), (preds * 255).astype('uint8')
 81 |             imgs1 = inputs[:, :3, :, :].transpose(0, 2, 3, 1)
 82 |             imgs2 = inputs[:, 3:6, :, :].transpose(0, 2, 3, 1)
 83 |             targets = targets.transpose(0, 2, 3, 1)
 84 |             preds = preds.transpose(0, 2, 3, 1)
 85 |             targets = targets[:, :, :, 0]
 86 |             preds = preds[:, :, :, 0]
 87 | 
 88 |             fig, axs = plt.subplots(imgs1.shape[0], 4)
 89 |             for i in range(imgs1.shape[0]):
 90 |                 if i == 0:
 91 |                     axs[i, 0].set_title('I_1')
 92 |                     axs[i, 1].set_title('I_2')
 93 |                     axs[i, 2].set_title('M')
 94 |                     axs[i, 3].set_title('P')
 95 |                 for j, imgs in enumerate([imgs1, imgs2, targets, preds]):
 96 |                     if j <= 1:
 97 |                         axs[i, j].imshow(imgs[i])
 98 |                     else:
 99 |                         axs[i, j].imshow(imgs[i], cmap='gray')
100 |                     axs[i, j].axis('off')
101 | 
102 |             fig.savefig(folder + '/{}.png'.format(epoch))
103 | 
104 | 
105 | 
106 | 
107 | 
108 | 


--------------------------------------------------------------------------------