为何我无法使用gpu运行

[Mayuyu007123]

app_info.py i changed to GPU, but still using my cpu to labeling

[CVHub520]

您好，可以参考下https://github.com/CVHub520/X-AnyLabeling/issues/188.

[Mayuyu007123]

i am using

the requirements-gpu.txt :

opencv-contrib-python-headless==4.7.0.72 PyQt5==5.15.7 imgviz==1.5.0 natsort==8.1.0 termcolor==1.1.0 PyYAML==6.0 onnx==1.13.1 onnxruntime-gpu>=1.17.0 qimage2ndarray==1.10.0 lapx==0.5.5 darkdetect==0.8.0 tqdm scipy shapely pyclipper filterpy tokenizers

but after pip install the pip install -r requirements-gpu-dev.txt

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. inference 0.9.11 requires onnxruntime<=1.15.1, which is not installed. autodistill-yolov8 0.1.4 requires ultralytics==8.0.81, but you have ultralytics 8.1.34 which is incompatible. inference 0.9.11 requires shapely<=2.0.1, but you have shapely 2.0.3 which is incompatible. inference-gpu 0.9.12rc1 requires inference-cli==0.9.12rc1, but you have inference-cli 0.9.11 which is incompatible. inference-gpu 0.9.12rc1 requires onnxruntime-gpu<=1.15.1, but you have onnxruntime-gpu 1.17.1 which is incompatible. inference-gpu 0.9.12rc1 requires shapely<=2.0.1, but you have shapely 2.0.3 which is incompatible.

[Mayuyu007123]

when i downgrade to the compatible version is same still using cpu

[Mayuyu007123]

您好，可以参考下#188. 我發現直接使用gpu的exe選擇 model的時候就會沒有回應然後應用程式直接跳出來,但是cpu的exe沒有

[CVHub520]

We highly recommend that you create a separate virtual environment and install the compatible version of the ORT-GPU package to avoid dependency conflicts.