S-Rajesh/triqa-iqa

1# Upload Instructions for Hugging Face Space2 3## Files to Upload4 5Upload these files to your Hugging Face Space (in this order):6 7### 1. Core Application Files8- `app.py` - Main Gradio application9- `README.md` - Space description10- `requirements.txt` - Python dependencies11- `package.json` - Node.js configuration12 13### 2. Model Architecture Files14- `convnext_original.py` - Original ConvNeXt model15- `convnext_finetune.py` - Fine-tuned ConvNeXt model16 17### 3. Sample Images18- Upload the entire `sample_image/` folder with all images19 20### 4. Model Weights (Download from Box first)21**IMPORTANT**: Download these files from Box first: https://utexas.box.com/s/8aw6axc2lofouja65uc726lca8b1cduf22 23Then create these directories and upload:24- `feature_models/convnext_tiny_22k_224.pth` (170MB)25- `feature_models/triqa_quality_aware.pth` (107MB)26- `Regression_Models/KonIQ_scaler.save`27- `Regression_Models/KonIQ_TRIQA.save` (111MB)28 29## Upload Steps30 311. **Create the Space** on Hugging Face322. **Upload files** using the web interface or Git333. **Wait for build** (2-5 minutes)344. **Test the demo** with sample images35 36## Expected Result37 38The demo will show:39- Image upload interface40- Quality assessment button41- Sample images for testing42- Quality score output (1-5 scale)43- Paper links and citation44 45## Troubleshooting46 47If the demo doesn't work:481. Check that all model files are uploaded492. Verify file paths are correct503. Check the Space logs for errors514. Ensure all dependencies are in requirements.txt52