This project is based on the Kaggle "2018 Data Science Bowl". It is implemented in Tensorflow 2. The main task is to make a model that can identify a range of nuclei across varied conditions. The results are displayed in the notebook. There are plenty of scopes for improvement. Major highlights of this project:
- Knowledge about medical data analysis using computer vision
- Implementation of datagenerator for semi supervised learning
- Implementation of U- Net architecture from scratch for image segmentation
- Optimization of the 'intersection over union' loss function to improve the performance of the model.