search

matterport / Mask_RCNN

Mask R-CNN for object detection and instance segmentation on Keras and TensorFlow
Other
24.72k stars 11.71k forks source link

GPU memory is occupied but cannot proceed #3044

Open r2race opened 4 months ago

r2race commented 4 months ago

GPU is fully occupied, as shown in the figure below,：

dd652e615e6ff64f59abfe21fd586b3

But training cannot proceed and is stuck at 1/1： image The code can proceed on cpu, but it's too slow，how can I solve this problem

Becerril42 commented 3 months ago

I'm not an expert in GPUs and Python, but doesn't the 0% below GPU-Util indicate that no GPU is being used? Are you using Windows or Linux? For Tensorflow 2.11 and above, Python can detect the GPU, but is unable to use it