Refer to the report for the explanation of the entire project and the segmentation.ipynb file for the code.
Results of the project so far:
Masked RCNN
Plots
Inferences
The Mask R-CNN model achieved the highest IoU scores for both the DSB-2018 dataset (0.91) and the SegPC-2021 dataset (0.89). This indicates that the model performed well in accurately segmenting nuclei in both datasets.
Unet++
Inferences
The UNet++ model achieved an IoU score of 0.85 on the DSB-2018 dataset and 0.81 on the SegPC-2021 dataset. This model performed better than the UNet model on both datasets, indicating that the improved architecture of UNet++ resulted in better performance.
Sample Ouput
