v1.0.0

🚀 Jetson-CV-Opt v1.0.0 Release Notes

This release provides all necessary resources for reproducing our computer vision model optimization pipeline on NVIDIA Jetson AGX Orin. Pre-converted ONNX models, configuration files, and INT8 calibration tables are included to facilitate high-throughput, NPU-accelerated inference with minimal accuracy loss.

---

🔹 What's Included

• Pre-trained ONNX models for key vision architectures (ConvMixer, EfficientFormerV2, FastViT, MobileOne, Res2Net, YOLOv9, etc.)
• Benchmark configuration files for classification and detection tasks
• INT8 calibration tables for mixed-precision deployment and DLA support
• Dataset folder structure guidance

---

🔹 Notes

• All ONNX models are exported with static input shapes and are compatible with TensorRT 10.7.
• INT8 calibration tables are required for optimal NPU/DLA inference.
• For custom models or retraining, please refer to the hyperparameter tables and training guides in our repository.

If you have questions or feedback, feel free to open an issue or pull request.

Full Changelog: https://github.com/cap-lab/Jetson-CV-Opt/commits/v1.0.0