Architectures and baseline models for the Nucleus Segmentation Challenge as part of the 2018 Data Science Bowl hosted by Kaggle.
-
Weighted U-Net/ - contains all scripts required to run weighted U-net model including preprocessing.
-
U-Net/ - contains all scripts required to run diff versions of u-net
-
Mask R-CNN/ - contains all scripts required to run mask r-cnn
-
Ensemble/ - contains all scripts for the ensemble models
-
Report/ - contains the report describing the architectures, preprocessing and results in detail
-
https://www.kaggle.com/keegil/keras-u-net-starter-lb-0-277?scriptVersionId=2164855 - used to construct baseline U-Net model, and is further modified to construct other versions of U-Net and is then used in ensembling U-Nets and U-Net and Mask R-CNN.
-
https://www.kaggle.com/aglotero/another-iou-metric - used to build evaluation metric to establish standard way of evaluating all our models.
-
https://github.com/matterport/Mask_RCNN/tree/master/samples - used as reference to construct MRCNN model to establish a baseline and then use in ensembling Mask R-CNN and U-Net
Please feel free to contact me at ananth360@gmail.com for any images/weights required to run the model or for permission to use the code or material in the report.