Summer@EPFL IVRL 2023
State-of-the-art segmentation investigation (Train, Evaluate) on an old comic dataset.
- src: contains main classes and functions
- pyscripts: contatins python scripts for training, evaluating on datasets
- scripts: contains bash scripts for running pyscripts files
Just fix the addresses, then put your bash script in 'scripts' folder.
sh scripts/mask2former_predict.sh
- If comic dataset is in cityscapes format, use convert_comic_to_coco_format.ipynb notebook, and then update the path of json_file in src.dataset.register_comic_instances.py
- If comic dataset is in coco format, all the src functions will be ok.
- All available codes can be found in src.train
- Both semantic & instance evaluation codes are in src.evaluation
- Functions can be found in src.dataset.helpers
For more info, check available notebooks in the notebooks directory.
Mean IoU on Placid comic:
Important Class | Character | Text | Comic Bubble |
---|---|---|---|
Pre-trained | 0.8980 | 0.9096 | 0.6771 |
Pre-trained (w/o sem_seg_head.class_embed) | 0.5235 | 0.2613 | 0.3351 |
Fine-tuned class-embed | 0.5803 | 0.5124 | 0.3844 |
Fine-tuned query-embeds | 0.5775 | 0.5083 | 0.3882 |
Fine-tuned decoder | 0.5711 | 0.3739 | 0.2643 |
Important Class | Character | Text | Comic Bubble |
---|---|---|---|
Pre-trained | 0.5576 | 0.5080 | 0.4104 |
Pre-trained (w/o sem_seg_head.predictor) | 0.3423 | 0.2677 | 0.1915 |
Fine-tuned predictor | 0.1097 | 0.0929 | 0.0735 |
Fine-tuned decoder | 0.1654 | 0.1423 | 0.1120 |
Fine-tuned whole model | 0.0908 | 0.0721 | 0.0584 |