search

AILab-CVC / YOLO-World

[CVPR 2024] Real-Time Open-Vocabulary Object Detection
https://www.yoloworld.cc
GNU General Public License v3.0
4.26k stars 416 forks source link

C++/libtorch implementation #261

Open RoboEvangelist opened 4 months ago

RoboEvangelist commented 4 months ago

Where can get a high performance version of the code, such the pytorch c++ version?

Thanks,

wondervictor commented 4 months ago

Hi @RoboEvangelist, the C++ version has not been developed but is on the TODO list. Currently, you can try to export ONNX models and adopt TensorRT or other acceleration tools.

jzx-gooner commented 4 months ago

Hi @RoboEvangelist, the C++ version has not been developed but is on the TODO list. Currently, you can try to export ONNX models and adopt TensorRT or other acceleration tools.

do we have a tensorrt demo？

PrinceP commented 1 month ago

@jzx-gooner https://github.com/PrinceP/tensorrt-cpp-for-onnx?tab=readme-ov-file#yolo-world

RoboEvangelist commented 1 month ago

@jzx-gooner https://github.com/PrinceP/tensorrt-cpp-for-onnx?tab=readme-ov-file#yolo-world

Thanks, broh. This is great!

RoboEvangelist commented 3 weeks ago

@jzx-gooner https://github.com/PrinceP/tensorrt-cpp-for-onnx?tab=readme-ov-file#yolo-world

@PrinceP What is the processing speed and memory usage compared with the python version?

Thanks

PrinceP commented 3 weeks ago

@RoboEvangelist For onnx model: Average time per image: 0.3813 seconds

For trt model: Average time per image: 0.0804 seconds