upload_instructions.md

Upload Instructions for Hugging Face Space

Files to Upload

Upload these files to your Hugging Face Space (in this order):

1. Core Application Files

• app.py - Main Gradio application
• README.md - Space description
• requirements.txt - Python dependencies
• package.json - Node.js configuration

2. Model Architecture Files

• convnext_original.py - Original ConvNeXt model
• convnext_finetune.py - Fine-tuned ConvNeXt model

3. Sample Images

• Upload the entire sample_image/ folder with all images

4. Model Weights (Download from Box first)

IMPORTANT: Download these files from Box first: https://utexas.box.com/s/8aw6axc2lofouja65uc726lca8b1cduf

Then create these directories and upload:

• feature_models/convnext_tiny_22k_224.pth (170MB)
• feature_models/triqa_quality_aware.pth (107MB)
• Regression_Models/KonIQ_scaler.save
• Regression_Models/KonIQ_TRIQA.save (111MB)

Upload Steps

1. Create the Space on Hugging Face
2. Upload files using the web interface or Git
3. Wait for build (2-5 minutes)
4. Test the demo with sample images

Expected Result

The demo will show:

• Image upload interface
• Quality assessment button
• Sample images for testing
• Quality score output (1-5 scale)
• Paper links and citation

Troubleshooting

If the demo doesn't work:

1. Check that all model files are uploaded
2. Verify file paths are correct
3. Check the Space logs for errors
4. Ensure all dependencies are in requirements.txt