AI Model Library

Browse our collection of forensic detection models — from local ONNX inference to cloud-powered AI. Download available models or suggest new ones for the pipeline.

Detection Models

Production-ready models included with Forensic Media Analyzer. Each model has been validated for accuracy and integrated into the analysis pipeline.

Loading available models...

Pipeline Models

Models under evaluation for future integration. These candidates are sourced from HuggingFace, GitHub, and academic research, and are scored for forensic relevance.

Loading pipeline models...

Model Types Explained

Forensic Media Analyzer uses three categories of AI models, each with distinct tradeoffs for speed, privacy, and capability.

ONNX Models

Optimized for fast, local inference. Run entirely on your machine with no internet required. Best for privacy-sensitive cases.

  • Fully offline, no data leaves your machine
  • CPU & GPU acceleration via ONNX Runtime
  • Consistent, reproducible results

PyTorch / Torch Models

Research-grade deep learning models including Noiseprint, ManTra-Net, and CNN Universal. Requires torch/transformers runtime.

  • State-of-the-art research architectures
  • GPU-accelerated for batch processing
  • Spatial manipulation localization

Cloud AI Models

Leverage powerful cloud APIs from OpenAI, Google, Anthropic, and others for multimodal analysis with comprehensive forensic prompts.

  • Largest, most capable models available
  • Natural language forensic reasoning
  • 6 providers for cross-validation

Suggest a Model

Know of a forensic detection model that should be in our library? Let us know and we will evaluate it for integration.