Releases: Cheng-Lin-Li/SegCaps
Releases · Cheng-Lin-Li/SegCaps
Version 1.1
- Include new pre-trained model and weight file.
- Enrich program documentations.
First release
Enhancements & Modifications
- The program was modified to support python 3.6 on Ubuntu 18.04 and Windows 10.
- Support not only 3D computed tomography scan images but also 2D Microsoft Common Objects in COntext (MS COCO) dataset images.
- Change the dice loss function type from Sørensen to Jaccard coefficient for comparing the similarity
- Add Kfold parameter for users to customize the cross validation task. K = 1 will force model to perform overfit.
- Add retrain parameter to enable users to reload pre-trained weights and retrain the model.
- Add initial learning rate for users to adjust.
- Add steps per epoch for users to adjust.
- Add number of patience for early stop of training to users.
- Add 'bce_dice' loss function as binary cross entropy + soft dice coefficient.
- Revise 'train', 'test', 'manip' flags from 0 or 1 to flags show up or not to indicate the behavior of main program.
- Add new webcam integration program for video stream segmentation