Prism is a state-of-the-art AI application designed for deep content analysis of images and documents. It leverages advanced multi-modal models to ensure content compliance, safety, and quality control.

• Ribbon & Layout Detection: precision detection of visual elements.
• Image Quality Assessment: Checks for blur, resolution, and pixel variance.
• GNC (General Non-Compliance) Checks: Identifies unauthorized visual gestures or icons.

• Tagline Verification: content and legibility analysis.
• Risk Assessment: Detects high-risk content, gambling references, and illegal activities.
• Competitor Analysis: Identifies competitor brand mentions.
• Sensitive Content Shield: Filters inappropriate, religious, or violent imagery.

• Single & Batch Processing: Analyze one image or thousands at once.
• Interactive Reports: Detailed breakdown of passes, fails, and specific issues.
• High-Performance Backbone: Powered by InternVL2.5-1B-MPO for vision-language understanding and EasyOCR for robust text extraction.

Experience Prism 2.0 live on Hugging Face Spaces: 👉 https://devranx-prism2-0.hf.space/

• Frontend: React (Vite) with a premium, responsive UI.
• Backend: Flask (Python) for API handling and orchestration.
• AI Models:
  • InternVL2.5-1B-MPO (Vision Language Model)
  • EasyOCR (Optical Character Recognition)
  • CLIP (Semantic Understanding)
• Deployment: Dockerized for seamless scalability.

1. Clone the repository

   git clone https://github.com/devsingh02/PRISM.git
   cd PRISM

2. Set up Python Environment

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

3. Run the Application

   python app.py

   The app will run at http://localhost:7860.

Prism is fully containerized.

docker build -t prism-app .
docker run -p 7860:7860 prism-app