Spaces:
Sleeping
Sleeping
|
Download README.md from Monishcloud8/mathsnap-deblur: direct link, hf CLI and curl.
- Browser
- Download file 941 Bytes
-
https://huggingface.co/spaces/Monishcloud8/mathsnap-deblur/resolve/main/README.md
- Command line
-
hf download hf://spaces/Monishcloud8/mathsnap-deblur/README.md
-
curl -L -o README.md https://huggingface.co/spaces/Monishcloud8/mathsnap-deblur/resolve/main/README.md
941 Bytes
| title: MathSnap Deblur | |
| emoji: 🔍 | |
| colorFrom: green | |
| colorTo: gray | |
| sdk: gradio | |
| sdk_version: 6.31.0 | |
| python_version: "3.12" | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| # MathSnap Deblur | |
| Sharpens blurry photos of printed math pages for the MathSnap app's preview. It runs NAFNet | |
| (GoPro motion-deblur, width 64) on free ZeroGPU hardware, in 384×384 tiles with feathered seams. | |
| - Endpoint: Gradio API `/deblur` taking `[jpeg_base64, key]`. It returns a base64 JPEG of at most 2048 px. | |
| - Only the MathSnap Supabase function knows the key. Set it as the Space secret `DEBLUR_KEY` (32+ characters). | |
| - Images are processed in memory and are never logged or stored. | |
| Model: [NAFNet](https://github.com/megvii-research/NAFNet) by megvii-research (MIT). Weights: | |
| [mlx-community/NAFNet-GoPro-width64](https://huggingface.co/mlx-community/NAFNet-GoPro-width64) (MIT), | |
| pinned to revision `09d01a2c` and checked against their SHA-256 at startup. | |