├── README.md
├── VGG_mean.mat
├── demo_a_fcn.m
├── demo_m_fcn.m
├── example.avi
├── example
    ├── flow_x_0001.jpg
    ├── flow_x_0002.jpg
    ├── flow_x_0003.jpg
    ├── flow_x_0004.jpg
    ├── flow_x_0005.jpg
    ├── flow_x_0006.jpg
    ├── flow_x_0007.jpg
    ├── flow_x_0008.jpg
    ├── flow_x_0009.jpg
    ├── flow_x_0010.jpg
    ├── flow_x_0011.jpg
    ├── flow_x_0012.jpg
    ├── flow_x_0013.jpg
    ├── flow_x_0014.jpg
    ├── flow_x_0015.jpg
    ├── flow_x_0016.jpg
    ├── flow_x_0017.jpg
    ├── flow_x_0018.jpg
    ├── flow_x_0019.jpg
    ├── flow_x_0020.jpg
    ├── flow_y_0001.jpg
    ├── flow_y_0002.jpg
    ├── flow_y_0003.jpg
    ├── flow_y_0004.jpg
    ├── flow_y_0005.jpg
    ├── flow_y_0006.jpg
    ├── flow_y_0007.jpg
    ├── flow_y_0008.jpg
    ├── flow_y_0009.jpg
    ├── flow_y_0010.jpg
    ├── flow_y_0011.jpg
    ├── flow_y_0012.jpg
    ├── flow_y_0013.jpg
    ├── flow_y_0014.jpg
    ├── flow_y_0015.jpg
    ├── flow_y_0016.jpg
    ├── flow_y_0017.jpg
    ├── flow_y_0018.jpg
    ├── flow_y_0019.jpg
    └── flow_y_0020.jpg
└── proto
    ├── actionness_a-fcn_scale_1_deploy.prototxt
    ├── actionness_a-fcn_scale_2_deploy.prototxt
    ├── actionness_a-fcn_scale_3_deploy.prototxt
    ├── actionness_a-fcn_scale_4_deploy.prototxt
    ├── actionness_m-fcn_scale_1_deploy.prototxt
    ├── actionness_m-fcn_scale_2_deploy.prototxt
    ├── actionness_m-fcn_scale_3_deploy.prototxt
    └── actionness_m-fcn_scale_4_deploy.prototxt


/README.md:
--------------------------------------------------------------------------------
 1 | # Actionness Estimation Using Hybrid FCNs
 2 | Here we provide the code of actionness estimation with hybrid fully convolutional networks from the following paper:
 3 | 
 4 |     Actionness Estimation Using Hybrid Fully Convolutional Netoworks
 5 |     Limin Wang, Yu Qiao, Xiaou Tang, and Luc Van Gool, in CVPR, 2016
 6 |     
 7 | ### Updates
 8 | 
 9 | - Jul 22, 2016
10 |   * Initilaize repo of actionness estimation.
11 | 
12 | ### Demo code
13 | - **demo_a_fcn.m**: an example showing actionness estimation with A-FCN.
14 | - **demo_m_fcn.m**: an example showing actionness estimation with M-FCN.
15 | - For optical flow extraction, we use TVL1 Optical Flow </br>
16 | You need download our dense flow code and compile it by yourself. [Dense Flow](https://github.com/wanglimin/dense_flow)
17 |   
18 | ### Download
19 | - Actionness estimation models (A-FCN) on the dataset of Stanford 40: <br />
20 |   http://mmlab.siat.ac.cn/actionness/stanford40_actionness_a-fcn.caffemodel
21 | - Actionness estimation models (A-FCN and M-FCN) on the dataset of UCF Sports: <br />
22 |   http://mmlab.siat.ac.cn/actionness/ucf_sports_actionness_a-fcn.caffemodel
23 |   http://mmlab.siat.ac.cn/actionness/ucf_sports_actionness_m-fcn.caffemodel
24 | - Actionness estimation models (A-FCN and M-FCN) on the dataset of JHMDB: <br />
25 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split1_actionness_a-fcn.caffemodel
26 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split1_actionness_m-fcn.caffemodel <br />
27 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split2_actionness_a-fcn.caffemodel
28 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split2_actionness_m-fcn.caffemodel <br />
29 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split3_actionness_a-fcn.caffemodel
30 |   http://mmlab.siat.ac.cn/actionness/jhmdb_split3_actionness_m-fcn.caffemodel
31 | 
32 | ### Questions
33 | Contact 
34 | - [Limin Wang](http://wanglimin.github.io/)
35 | 
36 | 
37 | 


--------------------------------------------------------------------------------
/VGG_mean.mat:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/VGG_mean.mat


--------------------------------------------------------------------------------
/demo_a_fcn.m:
--------------------------------------------------------------------------------
  1 | % demo code of actionness estimation
  2 | rgb_video =  'example.avi';
  3 | flow_video = 'example/';
  4 | gpu_id = 0;
  5 | 
  6 | 
  7 | % Read video
  8 | vidObj = VideoReader(rgb_video);
  9 | video = read(vidObj);
 10 | 
 11 | % Data preparation
 12 | IMAGE_MEAN = load('VGG_mean.mat');
 13 | IMAGE_MEAN = IMAGE_MEAN.image_mean;
 14 | IMAGE_MEAN = imresize(IMAGE_MEAN,[240,320]);
 15 | test_image = single(video(:,:,[3,2,1],:));
 16 | test_image = bsxfun(@minus,test_image,IMAGE_MEAN);
 17 | test_image = permute(test_image,[2,1,3,4]);
 18 | 
 19 | batch_size = 50;
 20 | num_images = size(test_image,4);
 21 | num_batches = ceil(num_images/batch_size);
 22 | 
 23 | model_file = 'jhmdb_split1_actionness_a-fcn.caffemodel';
 24 | 
 25 | % Multi-scale test
 26 | scale = 1;
 27 | model_def_file = ['proto/actionness_a-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 28 | caffe.reset_all();
 29 | caffe.set_mode_gpu();
 30 | caffe.set_device(gpu_id);
 31 | net = caffe.Net(model_def_file, model_file, 'test');
 32 | 
 33 | a_fcn_scale_1 = zeros(10, 13, 2, size(test_image,4));
 34 | images = zeros(214, 160, 3, batch_size, 'single');
 35 | for bb = 1 : num_batches
 36 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 37 |     tmp = test_image(:,:,:,range);
 38 |     for i  =1:size(tmp,4)
 39 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[214, 160]);
 40 |     end
 41 |     net.blobs('data').set_data(images);
 42 |     net.forward_prefilled();
 43 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 44 |     a_fcn_scale_1(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
 45 | end
 46 | 
 47 | scale = 2;
 48 | model_def_file = ['proto/actionness_a-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 49 | caffe.reset_all();
 50 | caffe.set_mode_gpu();
 51 | caffe.set_device(gpu_id);
 52 | net = caffe.Net(model_def_file, model_file, 'test');
 53 | 
 54 | a_fcn_scale_2 = zeros(15, 20, 2, size(test_image,4));
 55 | images = zeros(320, 240, 3, batch_size, 'single');
 56 | for bb = 1 : num_batches
 57 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 58 |     tmp = test_image(:,:,:,range);
 59 |     images(:,:,:,1:size(tmp,4)) = tmp;
 60 |     net.blobs('data').set_data(images);
 61 |     net.forward_prefilled();
 62 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 63 |     a_fcn_scale_2(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
 64 | end
 65 | 
 66 | scale = 3;
 67 | model_def_file = ['proto/actionness_a-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 68 | caffe.reset_all();
 69 | caffe.set_mode_gpu();
 70 | caffe.set_device(gpu_id);
 71 | net = caffe.Net(model_def_file, model_file, 'test');
 72 | 
 73 | a_fcn_scale_3 = zeros(22, 30, 2, size(test_image,4));
 74 | images = zeros(480, 360, 3, batch_size, 'single');
 75 | for bb = 1 : num_batches
 76 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 77 |     tmp = test_image(:,:,:,range);
 78 |     for i  =1:size(tmp,4)
 79 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[480, 360]);
 80 |     end
 81 |     net.blobs('data').set_data(images);
 82 |     net.forward_prefilled();
 83 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 84 |     a_fcn_scale_3(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
 85 | end
 86 | 
 87 | scale = 4;
 88 | model_def_file = ['proto/actionness_a-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 89 | caffe.reset_all();
 90 | caffe.set_mode_gpu();
 91 | caffe.set_device(gpu_id);
 92 | net = caffe.Net(model_def_file, model_file, 'test');
 93 | 
 94 | a_fcn_scale_4 = zeros(30, 40, 2, size(test_image,4));
 95 | images = zeros(640, 480, 3, batch_size, 'single');
 96 | for bb = 1 : num_batches
 97 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 98 |     tmp = test_image(:,:,:,range);
 99 |     for i  =1:size(tmp,4)
100 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[640,480]);
101 |     end
102 |     net.blobs('data').set_data(images);
103 |     net.forward_prefilled();
104 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
105 |     a_fcn_scale_4(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
106 | end
107 | 
108 | 
109 | for i = 1:size(video);
110 |     subplot(1,2,1);
111 |     imshow(video(:,:,:,i));
112 |     subplot(1,2,2);
113 |     result = (imresize(a_fcn_scale_1(:,:,2,i),[240,320]) ...
114 |         +imresize(a_fcn_scale_2(:,:,2,i),[240,320])...
115 |         +imresize(a_fcn_scale_3(:,:,2,i),[240,320])...
116 |         +imresize(a_fcn_scale_4(:,:,2,i),[240,320]))/4;
117 |     imagesc(result);
118 |     axis image; axis off;
119 |     pause(1);
120 | end


--------------------------------------------------------------------------------
/demo_m_fcn.m:
--------------------------------------------------------------------------------
  1 | % demo code of actionness estimation
  2 | path_flow = 'example/';
  3 | gpu_id = 0;
  4 | filelist = dir([path_flow, '*.jpg']);
  5 | duration = length(filelist)/2 + 1;
  6 | 
  7 | 
  8 | % Read optical flow
  9 | flow = zeros(240, 320, 4, duration);
 10 | pre_flow_x = []; cur_flow_x = [];
 11 | pre_flow_y = []; cur_flow_y = [];
 12 | for k = 1:duration
 13 |     if k < duration
 14 |         name_x = sprintf('flow_x_%04d.jpg',k);
 15 |         name_y = sprintf('flow_y_%04d.jpg',k);
 16 |         if isempty(pre_flow_x)
 17 |             pre_flow_x = imresize(imread([path_flow,'/',name_x]),[240,320]);
 18 |             pre_flow_y = imresize(imread([path_flow,'/',name_y]),[240,320]);
 19 |         end
 20 |         cur_flow_x = imresize(imread([path_flow,'/',name_x]),[240,320]);
 21 |         cur_flow_y = imresize(imread([path_flow,'/',name_y]),[240,320]);
 22 |     end
 23 |     flow(:,:,:,k) = cat(3,pre_flow_x,pre_flow_y,cur_flow_x,cur_flow_y);
 24 |     pre_flow_x = cur_flow_x;
 25 |     pre_flow_y = cur_flow_y;
 26 | end
 27 | 
 28 | 
 29 | 
 30 | % Data preparation
 31 | flow(:) = flow(:) -128;
 32 | test_image = permute(flow,[2,1,3,4]);
 33 | 
 34 | batch_size = 50;
 35 | num_images = size(test_image,4);
 36 | num_batches = ceil(num_images/batch_size);
 37 | 
 38 | model_file = 'jhmdb_split1_actionness_m-fcn.caffemodel';
 39 | 
 40 | % Multi-scale test
 41 | scale = 1;
 42 | model_def_file = ['proto/actionness_m-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 43 | caffe.reset_all();
 44 | caffe.set_mode_gpu();
 45 | caffe.set_device(gpu_id);
 46 | net = caffe.Net(model_def_file, model_file, 'test');
 47 | 
 48 | m_fcn_scale_1 = zeros(10, 13, 2, size(test_image,4));
 49 | images = zeros(214, 160, 4, batch_size, 'single');
 50 | for bb = 1 : num_batches
 51 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 52 |     tmp = test_image(:,:,:,range);
 53 |     for i  =1:size(tmp,4)
 54 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[214, 160]);
 55 |     end
 56 |     net.blobs('data').set_data(images);
 57 |     net.forward_prefilled();
 58 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 59 |     m_fcn_scale_1(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
 60 | end
 61 | 
 62 | scale = 2;
 63 | model_def_file = ['proto/actionness_m-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 64 | caffe.reset_all();
 65 | caffe.set_mode_gpu();
 66 | caffe.set_device(gpu_id);
 67 | net = caffe.Net(model_def_file, model_file, 'test');
 68 | 
 69 | m_fcn_scale_2 = zeros(15, 20, 2, size(test_image,4));
 70 | images = zeros(320, 240, 4, batch_size, 'single');
 71 | for bb = 1 : num_batches
 72 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 73 |     tmp = test_image(:,:,:,range);
 74 |     images(:,:,:,1:size(tmp,4)) = tmp;
 75 |     net.blobs('data').set_data(images);
 76 |     net.forward_prefilled();
 77 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 78 |     m_fcn_scale_2(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
 79 | end
 80 | 
 81 | scale = 3;
 82 | model_def_file = ['proto/actionness_m-fcn_scale_', num2str(scale), '_deploy.prototxt'];
 83 | caffe.reset_all();
 84 | caffe.set_mode_gpu();
 85 | caffe.set_device(gpu_id);
 86 | net = caffe.Net(model_def_file, model_file, 'test');
 87 | 
 88 | m_fcn_scale_3 = zeros(22, 30, 2, size(test_image,4));
 89 | images = zeros(480, 360, 4, batch_size, 'single');
 90 | for bb = 1 : num_batches
 91 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
 92 |     tmp = test_image(:,:,:,range);
 93 |     for i  =1:size(tmp,4)
 94 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[480, 360]);
 95 |     end
 96 |     net.blobs('data').set_data(images);
 97 |     net.forward_prefilled();
 98 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
 99 |     m_fcn_scale_3(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
100 | end
101 | 
102 | scale = 4;
103 | model_def_file = ['proto/actionness_m-fcn_scale_', num2str(scale), '_deploy.prototxt'];
104 | caffe.reset_all();
105 | caffe.set_mode_gpu();
106 | caffe.set_device(gpu_id);
107 | net = caffe.Net(model_def_file, model_file, 'test');
108 | 
109 | m_fcn_scale_4 = zeros(30, 40, 2, size(test_image,4));
110 | images = zeros(640, 480, 4, batch_size, 'single');
111 | for bb = 1 : num_batches
112 |     range = 1 + batch_size*(bb-1): min(num_images,batch_size*bb);
113 |     tmp = test_image(:,:,:,range);
114 |     for i  =1:size(tmp,4)
115 |         images(:,:,:,i) = imresize(tmp(:,:,:,i),[640,480]);
116 |     end
117 |     net.blobs('data').set_data(images);
118 |     net.forward_prefilled();
119 |     prediction = permute(net.blobs('prob').get_data(), [2,1,3,4]);
120 |     m_fcn_scale_4(:,:,:,range) = prediction(:,:,:,mod(range-1,batch_size)+1);
121 | end
122 | 
123 | 
124 | for i = 1:size(video,4);
125 |     subplot(1,2,1);
126 |     imshow(video(:,:,:,i));
127 |     subplot(1,2,2);
128 |     result_a = (imresize(a_fcn_scale_1(:,:,2,i),[240,320]) ...
129 |         +imresize(a_fcn_scale_2(:,:,2,i),[240,320])...
130 |         +imresize(a_fcn_scale_3(:,:,2,i),[240,320])...
131 |         +imresize(a_fcn_scale_4(:,:,2,i),[240,320]))/4;
132 |     result_m = (imresize(m_fcn_scale_1(:,:,2,i),[240,320]) ...
133 |         +imresize(m_fcn_scale_2(:,:,2,i),[240,320])...
134 |         +imresize(m_fcn_scale_3(:,:,2,i),[240,320])...
135 |         +imresize(m_fcn_scale_4(:,:,2,i),[240,320]))/4;
136 |     imagesc(result_a + result_m);
137 |     axis image; axis off;
138 |     pause(1);
139 | end
140 | 


--------------------------------------------------------------------------------
/example.avi:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example.avi


--------------------------------------------------------------------------------
/example/flow_x_0001.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0001.jpg


--------------------------------------------------------------------------------
/example/flow_x_0002.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0002.jpg


--------------------------------------------------------------------------------
/example/flow_x_0003.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0003.jpg


--------------------------------------------------------------------------------
/example/flow_x_0004.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0004.jpg


--------------------------------------------------------------------------------
/example/flow_x_0005.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0005.jpg


--------------------------------------------------------------------------------
/example/flow_x_0006.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0006.jpg


--------------------------------------------------------------------------------
/example/flow_x_0007.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0007.jpg


--------------------------------------------------------------------------------
/example/flow_x_0008.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0008.jpg


--------------------------------------------------------------------------------
/example/flow_x_0009.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0009.jpg


--------------------------------------------------------------------------------
/example/flow_x_0010.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0010.jpg


--------------------------------------------------------------------------------
/example/flow_x_0011.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0011.jpg


--------------------------------------------------------------------------------
/example/flow_x_0012.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0012.jpg


--------------------------------------------------------------------------------
/example/flow_x_0013.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0013.jpg


--------------------------------------------------------------------------------
/example/flow_x_0014.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0014.jpg


--------------------------------------------------------------------------------
/example/flow_x_0015.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0015.jpg


--------------------------------------------------------------------------------
/example/flow_x_0016.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0016.jpg


--------------------------------------------------------------------------------
/example/flow_x_0017.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0017.jpg


--------------------------------------------------------------------------------
/example/flow_x_0018.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0018.jpg


--------------------------------------------------------------------------------
/example/flow_x_0019.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0019.jpg


--------------------------------------------------------------------------------
/example/flow_x_0020.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_x_0020.jpg


--------------------------------------------------------------------------------
/example/flow_y_0001.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0001.jpg


--------------------------------------------------------------------------------
/example/flow_y_0002.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0002.jpg


--------------------------------------------------------------------------------
/example/flow_y_0003.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0003.jpg


--------------------------------------------------------------------------------
/example/flow_y_0004.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0004.jpg


--------------------------------------------------------------------------------
/example/flow_y_0005.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0005.jpg


--------------------------------------------------------------------------------
/example/flow_y_0006.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0006.jpg


--------------------------------------------------------------------------------
/example/flow_y_0007.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0007.jpg


--------------------------------------------------------------------------------
/example/flow_y_0008.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0008.jpg


--------------------------------------------------------------------------------
/example/flow_y_0009.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0009.jpg


--------------------------------------------------------------------------------
/example/flow_y_0010.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0010.jpg


--------------------------------------------------------------------------------
/example/flow_y_0011.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0011.jpg


--------------------------------------------------------------------------------
/example/flow_y_0012.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0012.jpg


--------------------------------------------------------------------------------
/example/flow_y_0013.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0013.jpg


--------------------------------------------------------------------------------
/example/flow_y_0014.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0014.jpg


--------------------------------------------------------------------------------
/example/flow_y_0015.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0015.jpg


--------------------------------------------------------------------------------
/example/flow_y_0016.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0016.jpg


--------------------------------------------------------------------------------
/example/flow_y_0017.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0017.jpg


--------------------------------------------------------------------------------
/example/flow_y_0018.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0018.jpg


--------------------------------------------------------------------------------
/example/flow_y_0019.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0019.jpg


--------------------------------------------------------------------------------
/example/flow_y_0020.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/wanglimin/Actionness-Estimation/35221f84b765ee8b725f30648911df45b3788b54/example/flow_y_0020.jpg


--------------------------------------------------------------------------------
/proto/actionness_a-fcn_scale_1_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "A-FCN_Scale_1"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 3
  5 | input_dim: 160
  6 | input_dim: 214
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_a-fcn_scale_2_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "A-FCN_Scale_2"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 3
  5 | input_dim: 240
  6 | input_dim: 320
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4          
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_a-fcn_scale_3_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "A-FCN_Scale_3"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 3
  5 | input_dim: 360
  6 | input_dim: 480
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_a-fcn_scale_4_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "A-FCN_Scale_4"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 3
  5 | input_dim: 480
  6 | input_dim: 640
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_m-fcn_scale_1_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "M-FCN_Scale_1"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 4
  5 | input_dim: 160
  6 | input_dim: 214
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_m-fcn_scale_2_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "M-FCN_Scale_2"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 4
  5 | input_dim: 240
  6 | input_dim: 320
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4          
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_m-fcn_scale_3_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "M-FCN_Scale_4"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 4
  5 | input_dim: 360
  6 | input_dim: 480
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


--------------------------------------------------------------------------------
/proto/actionness_m-fcn_scale_4_deploy.prototxt:
--------------------------------------------------------------------------------
  1 | name: "M-FCN_Scale_4"
  2 | input: "data"
  3 | input_dim: 50
  4 | input_dim: 4
  5 | input_dim: 480
  6 | input_dim: 640
  7 | layer {
  8 |   name: "conv1"
  9 |   type: "Convolution"
 10 |   bottom: "data"
 11 |   top: "conv1"
 12 |   convolution_param {
 13 |     num_output: 96
 14 |     pad: 3
 15 |     kernel_size: 7
 16 |     stride: 2
 17 |   }
 18 | }
 19 | layer {
 20 |   name: "relu1"
 21 |   type: "ReLU"
 22 |   bottom: "conv1"
 23 |   top: "conv1"
 24 | }
 25 | layer {
 26 |   name: "norm1"
 27 |   type: "LRN"
 28 |   bottom: "conv1"
 29 |   top: "norm1"
 30 |   lrn_param {
 31 |     local_size: 5
 32 |     alpha: 0.0005
 33 |     beta: 0.75
 34 |   }
 35 | }
 36 | layer {
 37 |   name: "pool1"
 38 |   type: "Pooling"
 39 |   bottom: "norm1"
 40 |   top: "pool1"
 41 |   pooling_param {
 42 |     pool: MAX
 43 |     kernel_size: 3
 44 |     stride: 2
 45 |   }
 46 | }
 47 | layer {
 48 |   name: "conv2"
 49 |   type: "Convolution"
 50 |   bottom: "pool1"
 51 |   top: "conv2"
 52 |   convolution_param {
 53 |     num_output: 256
 54 |     pad: 2
 55 |     kernel_size: 5
 56 |     stride: 2
 57 |   }
 58 | }
 59 | layer {
 60 |   name: "relu2"
 61 |   type: "ReLU"
 62 |   bottom: "conv2"
 63 |   top: "conv2"
 64 | }
 65 | layer {
 66 |   name: "norm2"
 67 |   type: "LRN"
 68 |   bottom: "conv2"
 69 |   top: "norm2"
 70 |   lrn_param {
 71 |     local_size: 5
 72 |     alpha: 0.0005
 73 |     beta: 0.75
 74 |   }
 75 | }
 76 | layer {
 77 |   name: "pool2"
 78 |   type: "Pooling"
 79 |   bottom: "norm2"
 80 |   top: "pool2"
 81 |   pooling_param {
 82 |     pool: MAX
 83 |     kernel_size: 3
 84 |     stride: 2
 85 |   }
 86 | }
 87 | layer {
 88 |   name: "conv3"
 89 |   type: "Convolution"
 90 |   bottom: "pool2"
 91 |   top: "conv3"
 92 |   convolution_param {
 93 |     num_output: 512
 94 |     pad: 1
 95 |     kernel_size: 3
 96 |   }
 97 | }
 98 | layer {
 99 |   name: "relu3"
100 |   type: "ReLU"
101 |   bottom: "conv3"
102 |   top: "conv3"
103 | }
104 | layer {
105 |   name: "conv4"
106 |   type: "Convolution"
107 |   bottom: "conv3"
108 |   top: "conv4"
109 |   convolution_param {
110 |     num_output: 512
111 |     pad: 1
112 |     kernel_size: 3
113 |   }
114 | }
115 | layer {
116 |   name: "relu4"
117 |   type: "ReLU"
118 |   bottom: "conv4"
119 |   top: "conv4"
120 | }
121 | layer {
122 |   name: "conv5"
123 |   type: "Convolution"
124 |   bottom: "conv4"
125 |   top: "conv5"
126 |   convolution_param {
127 |     num_output: 512
128 |     pad: 1
129 |     kernel_size: 3
130 |   }
131 | }
132 | layer {
133 |   name: "relu5"
134 |   type: "ReLU"
135 |   bottom: "conv5"
136 |   top: "conv5"
137 | }
138 | layer {
139 |   name: "pool5"
140 |   type: "Pooling"
141 |   bottom: "conv5"
142 |   top: "pool5"
143 |   pooling_param {
144 |     pool: MAX
145 |     kernel_size: 3
146 |     stride: 1
147 |     pad: 1
148 |   }
149 | }
150 | layer {
151 |   name: "conv6"
152 |   type: "Convolution"
153 |   bottom: "pool5"
154 |   top: "conv6"
155 |   convolution_param {
156 |     num_output: 1024
157 |     kernel_size: 1
158 |   }
159 | }
160 | layer {
161 |   name: "relu6"
162 |   type: "ReLU"
163 |   bottom: "conv6"
164 |   top: "conv6"
165 | }
166 | layer {
167 |   name: "drop6"
168 |   type: "Dropout"
169 |   bottom: "conv6"
170 |   top: "conv6"
171 |   dropout_param {
172 |     dropout_ratio: 0.4
173 |   }
174 | }
175 | layer {
176 |   name: "conv7"
177 |   type: "Convolution"
178 |   bottom: "conv6"
179 |   top: "conv7"
180 |   convolution_param {
181 |     num_output: 512
182 |     kernel_size: 1
183 |   }
184 | }
185 | layer {
186 |   name: "relu7"
187 |   type: "ReLU"
188 |   bottom: "conv7"
189 |   top: "conv7"
190 | }
191 | layer {
192 |   name: "drop7"
193 |   type: "Dropout"
194 |   bottom: "conv7"
195 |   top: "conv7"
196 |   dropout_param {
197 |     dropout_ratio: 0.4
198 |   }
199 | }
200 | layer {
201 |   name: "conv8"
202 |   type: "Convolution"
203 |   bottom: "conv7"
204 |   top: "conv8"
205 |   convolution_param {
206 |     num_output: 2
207 |     kernel_size: 1
208 |   }
209 | }
210 | layer {
211 |   name: "prob"
212 |   type: "Softmax"
213 |   bottom: "conv8"
214 |   top: "prob"
215 | }


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