#4343rd of 80 in Open models
pinto_model_zoo
Pre-converted vision models across TFLite, ONNX, OpenVINO, CoreML and more
- Stars
- 4.6k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Nov 2023
- Language
- Python
4,109 on 22 Apr 20264,638 on 11 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
A repository of models converted between frameworks, mostly image classification and other vision models such as face, person re-identification, gaze and attribute recognition. Formats include TensorFlow, TFLite (FP32/FP16/INT8), EdgeTPU, OpenVINO IR, CoreML, TF.js, TF-TRT and ONNX. The README links to the author's conversion tools, such as onnx2tf, and ONNX processing utilities.
Who it is for: Edge and embedded developers needing converted or quantized vision models
Strengths
- Same model offered in many formats: TFLite, EdgeTPU, OpenVINO, CoreML, TF.js, TF-TRT, ONNX
- Table lists per-model quantization variants (FP16, INT8, dynamic range, weight) and input sizes
- Links to the author's conversion tools like onnx2tf and ONNX editing utilities
- Edge-oriented models, including EdgeTPU and Raspberry Pi targets
Weaknesses
- Each model folder has its own license; source model terms apply, not MIT
- Author states sample code to verify operation is largely skipped
- Last release is 2023-11-30; repo has no tagged releases since
- Early notes say MobileNetV3 and full integer quantization models do not return correctly
What it needs
- Models: TensorFlow, TFLite, EdgeTPU, OpenVINO, CoreML
What changed
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in Open models
See all 80| Rank | Project | Score |
|---|---|---|
| 1 | ruviewWiFi CSI sensing on ESP32 for presence, breathing, heart rate and pose | 94 out of 100 |
| 2 | layaSingle-pass multilingual classifier returning typed decisions with calibrated odds | 88 out of 100 |
| 3 | rf-detrReal-time transformer models for object detection, segmentation and keypoints | 81 out of 100 |
| #4 | timesfmPretrained time-series foundation model for zero-shot forecasting from Google Research | 78 out of 100 |
| #5 | open_clipPyTorch library for training and running CLIP-style image-text models | 78 out of 100 |