This repository extends the YOLOv10 model with additional output layers: num_detections, detection_boxes, detection_scores, and detection_classes. It also includes optimizations for exporting models with dynamic size and dynamic batch size, ensuring that the ONNX models are optimized for TensorRT builds.
This repository provides a practical implementation of YOLOv10 within DeepStream, a powerful tool for video analytics. Users can seamlessly integrate the YOLOv10 model into their projects using DeepStream, leveraging its capabilities for efficient video processing and analysis.
Just sharing!!
Repository 1: YOLOv10 with Enhanced Outputs and TensorRT Optimization
YOLOv10 Export ONNX
This repository extends the YOLOv10 model with additional output layers:![image](https://github.com/THU-MIG/yolov10/assets/22964932/ccff2a15-0f38-4a58-a57b-a66a10626af6)
num_detections
,detection_boxes
,detection_scores
, anddetection_classes
. It also includes optimizations for exporting models with dynamic size and dynamic batch size, ensuring that the ONNX models are optimized for TensorRT builds.Repository 2: DeepStream Integration for YOLOv10
DeepStream YOLOv10
This repository provides a practical implementation of YOLOv10 within DeepStream, a powerful tool for video analytics. Users can seamlessly integrate the YOLOv10 model into their projects using DeepStream, leveraging its capabilities for efficient video processing and analysis.