ivanessi/mvtec-reconseg-demo

MVTec AD Bottle + Wood ReconSeg Demo

Gradio demo for a reconstruction-segmentation anomaly detection model.

The app expects the checkpoint at:

checkpoints/reconseg_resnet50_bottlelong_woodhead_best.pt

The checkpoint is intentionally included in the Space through Git LFS.

Upload to Hugging Face Space with the full checkpoint

Option 1: upload through the Hugging Face Hub API. This does not require local git-lfs.

cd hf_app
python -m pip install -U huggingface_hub
huggingface-cli login
python upload_to_hf.py --repo-id <username>/<space-name>

Option 2: upload with Git LFS.

cd hf_app
git init
git lfs install
git lfs track "*.pt"
git add .
git commit -m "Add Gradio ReconSeg demo"
git branch -M main
git remote add origin https://huggingface.co/spaces/<username>/<space-name>
git push -u origin main

The model supports two categories: bottle and wood.