├── FindingNuclei[KaggleCompetition].html ├── README.md ├── RetinalDiseaseClassification.html ├── Task1_CountingPoints.html └── Task2_CraterDetector.pdf /README.md: -------------------------------------------------------------------------------- 1 | # ImageProcessing_AI 2 | This repository is based on projects developed during the "AI Image Processing" course. As classes were taught in Spanish some of the explanations or parts of code may appear in Spanish. 3 | 4 | # Repository Structure 5 | This repository is organized as following: 6 | - **Task1_CountingPoints**: 7 | - The objective of this task was to count the number of circles in a given image. In this activity, we implemented four aspects: 1) Image reading and conversion to grayscale 2) Thresholding 3) Mathematical 8 | morphology 4) Object detection and measurement. 9 | 10 | - **Task2_CraterDetector**: 11 | - For this activity, we were given an external Kaggle code (https://www.kaggle.com/code/benmanor/crater-object-detection-using-faster-rcnn), which consisted of a crater detection model. Our task was to make it work for the set of images from the "Martian/Lunar Crater Detection Dataset", also available on Kaggle(https://www.kaggle.com/datasets/lincolnzh/martianlunar-crater-detection-dataset/data). This database comprises images of craters on the moon and Mars. Therefore, the ultimate goal was to create a crater detection model specifically tailored for lunar and Martian craters. Additionally, we were asked to employ a different detection model than the one used in the provided external code. 12 | 13 | - **FindingNuclei[KaggleCompetition]**: 14 | - In the 2018 Kaggle Data Science Bowl competition, participants were challenged to create algorithms capable of automating the detection of cell nuclei. The underlying mission was to accelerate the development 15 | of cures for various diseases. The database consisted of images of segmented cell nuclei, making the algorithm essentially a segmentation algorithm. 16 | 17 | Our teacher used this competition as a basis to create a similar one in class. In this classroom competition, participants were provided with the pre-solved problem code (the algorithm was already created), and 18 | their task was to modify parts of the code to achieve the best score according to predefined evaluation metrics. This evaluation was performed on three randomly selected images out of the seven provided in the 19 | competition. 20 | 21 | - **RetinalDiseaseClassification**: 22 | - For our final project, we were tasked with creating an algorithm to detect, through retinal images, whether a person is suffering from a retinal disease or not. The model was developed by using the dataset provided by the "Retinal Disease Classification" from Kaggle: https://www.kaggle.com/datasets/andrewmvd/retinal-disease-classification. 23 | 24 | -------------------------------------------------------------------------------- /Task2_CraterDetector.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sandra-pinel/Projects_ImageProcessing_AI/858dca30f8bf2091a5953f55e84d83bf800c1b61/Task2_CraterDetector.pdf --------------------------------------------------------------------------------