export to ONNX with NMS

[Egorundel]

Does the export to ONNX work with the NMS module? And dynamic batch size. If so, how do I do it?

As I understand it, the NMS module only works for TF models?

[WongKinYiu]

#79 (comment)

[zahidzqj]

Why does an error occur when the onnx file generated by export.py is converted to trt_engine, and #79 (comment) is normal.

[huihui308]

Why does an error occur when the onnx file generated by export.py is converted to trt_engine, and #79 (comment) is normal.

What tensorrt's version you are using?

[zahidzqj]

Why does an error occur when the onnx file generated by export.py is converted to trt_engine, and #79 (comment) is normal.

What tensorrt's version you are using?

tensorrt==7.2.1.6

[laugh12321]

Hello everyone!

I would like to introduce my open-source project - TensoRT-YOLO, a tool for deploying YOLO Series (Support YOLOv9) with Efficient NMS in TensorRT.

Key Features

• Supports FLOAT32, FLOAT16 ONNX export, and TensorRT inference
• Supports YOLOv5, YOLOv8, YOLOv9, PP-YOLOE, and PP-YOLOE+
• Integrates EfficientNMS TensorRT plugin for accelerated post-processing
• Utilizes CUDA kernel functions to accelerate preprocess
• Supports C++ and Python inference

Perfomance

Perfomance Test using GPU RTX 2080Ti 22GB on AMD Ryzen 7 5700X 8-Core/ 128GB RAM.

Model Performance Evaluation using TensorRT engine using TensoRT-YOLO.

All models were deployed using FP16, BatchSize 4 and size 640.

YOLOv9 Series

This includes the YOLOv9-C, YOLOv9-E, YOLOv9-C-Converted, YOLOv9-E-Converted, GELAN-C and GELAN-E.

YOLOv9-C	YOLOv9-E	YOLOv9-C-Converted	YOLOv9-E-Converted	GELAN-C	GELAN-E
Average Latency: 36.615ms	Average Latency: 59.736ms	Average Latency: 19.689ms	Average Latency: 53.144ms	Average Latency: 19.557ms	Average Latency: 53.575ms

YOLOv8 Series

This includes the YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l and YOLOv8x.

YOLOv8n	YOLOv8s	YOLOv8m	YOLOv8l	YOLOv8x
Average Latency: 10.289ms	Average Latency: 12.459ms	Average Latency: 18.514ms	Average Latency: 24.926ms	Average Latency: 34.587ms