Pretrained Weights

PyTorch checkpoints for the MSDCT-UNet local motion-blur segmentation model (MultiTransformUNet architecture with 57-channel HiFST DCT features).

File	Train source	Size
`bocchi.pth`	BOCCHI	113 MB
`reloblur.pth`	ReLoBlur	113 MB
`blurdataset.pth`	BlurDataset	113 MB

Each file stores {"model_state": state_dict} only (no optimizer state). Total parameters per checkpoint: 29.45 M.

Download

pip install huggingface_hub
huggingface-cli download aianonymous12/msdct-unet \
    bocchi.pth reloblur.pth blurdataset.pth \
    --local-dir checkpoints

Companion data

Pre-built feature PKLs + raw img+mask are at https://huggingface.co/datasets/aianonymous12/BOCCHI.

License

CC BY-NC 4.0 — non-commercial research use.

Downloads last month: -; Downloads are not tracked for this model. How to track