QUICK_START.md

TRIQA Hugging Face Demo - Quick Start Guide

🚀 Create Your Hugging Face Space

Step 1: Create the Space

1. Go to https://huggingface.co/spaces
2. Click "Create new Space"
3. Fill in:
   • Space name: triqa-image-quality-assessment
   • License: MIT
   • SDK: Gradio
   • Hardware: CPU (or GPU if available)
   • Visibility: Public

Step 2: Upload Files

Upload these files from this folder to your Hugging Face Space:

Core Files (Upload these first):

• app.py - Main application
• README.md - Space description
• requirements.txt - Dependencies
• package.json - Node.js config
• convnext_original.py - Model architecture
• convnext_finetune.py - Model architecture
• sample_image/ folder - Sample images

Model Files (Download from Box first):

1. Go to: https://utexas.box.com/s/8aw6axc2lofouja65uc726lca8b1cduf
2. Download the model files
3. Create these directories in your Space:
   • feature_models/
   • Regression_Models/
4. Upload the model files to their respective directories

Step 3: Test the Demo

1. Wait for the build to complete (2-5 minutes)
2. Upload a sample image
3. Click "Assess Quality"
4. Verify you get a quality score (1-5 scale)

📁 File Structure in Your Space

your-space/
├── app.py
├── README.md
├── requirements.txt
├── package.json
├── convnext_original.py
├── convnext_finetune.py
├── sample_image/
│   ├── 233045618.jpg
│   ├── 25239707.jpg
│   ├── 44009500.jpg
│   ├── 5129172.jpg
│   └── 85119046.jpg
├── feature_models/
│   ├── convnext_tiny_22k_224.pth
│   └── triqa_quality_aware.pth
└── Regression_Models/
    ├── KonIQ_scaler.save
    └── KonIQ_TRIQA.save

🎯 Expected Demo Features

• Image Upload: Drag & drop or click to upload
• Quality Assessment: Click button to get score
• Sample Images: Pre-loaded examples
• Paper Links: arXiv and IEEE links
• Citation: BibTeX citation

🔧 Troubleshooting

• Build fails: Check requirements.txt
• Models not loading: Verify model files are uploaded
• File not found: Check file paths and names
• Memory issues: Use CPU hardware

📞 Support

• GitHub: https://github.com/rajeshsureddi/triqa
• Paper: https://arxiv.org/pdf/2507.12687

Your demo will be live at: https://huggingface.co/spaces/your-username/triqa-image-quality-assessment