PINTO0309/PINTO_model_zoo

A repository for storing models that have been inter-converted between various frameworks. Supported frameworks are TensorFlow, PyTorch, ONNX, OpenVINO, TFJS, TFTRT, TensorFlowLite (Float32/16/INT8), EdgeTPU, CoreML.

What it solves

Converting AI models between different software frameworks (like PyTorch to TensorFlow) and optimizing them for specific hardware (like mobile phones or Edge TPUs) is often a complex, manual process. This project provides a collection of pre-converted and pre-quantized models to save developers from doing this work themselves.

How it works

The repository acts as a "model zoo," storing a wide variety of neural network models that have been inter-converted across multiple formats. It supports a broad range of frameworks and runtimes, including:

  • TensorFlow / TensorFlow Lite (including Float32, Float16, and INT8 quantization)
  • PyTorch
  • ONNX
  • OpenVINO
  • CoreML
  • TFJS (TensorFlow.js)
  • EdgeTPU

Who it’s for

Developers and researchers who need to deploy existing AI models to edge devices, mobile applications, or specific hardware accelerators without having to write their own conversion and quantization scripts.

Highlights

  • Extensive Format Support: Models are available in formats like .tflite, .h5, .pb, .onnx, and .mlmodel.
  • Quantization Options: Provides models with various quantization levels (Weight, Integer, Full Integer, and Float16) to balance performance and accuracy.
  • Diverse Model Library: Includes a vast array of models for image classification and 2D object detection (e.g., EfficientNet, MobileNet, YOLO variants).
  • Hardware Optimized: Specifically includes models optimized for the EdgeTPU and Raspberry Pi.

Related

  • Project
    onnx/models

    A curated collection of pre-trained, state-of-the-art machine learning models in the ONNX format, enabling cross-framework compatibility and deployment.

  • Project
    onnx/onnxmltools

    A toolkit for converting machine learning models from various frameworks like TensorFlow, scikit-learn, and XGBoost into the ONNX format for cross-platform deployment.

  • Project
    onnx/tensorflow-onnx

    A tool to convert TensorFlow, Keras, TensorFlow.js, and TFLite models to the ONNX format for cross-platform compatibility and deployment.

  • Project
    qualcomm/ai-hub-models

    A collection of state-of-the-art ML models optimized for Qualcomm devices, featuring tools to compile, quantize, and profile models for on-device deployment.

  • Project
    microsoft/Olive

    An AI model optimization toolkit that composes the best optimization techniques to create efficient ONNX models for cloud or edge inference based on hardware and performance constraints.