taokong/RON

How to reduce batchsize when im testing ?

wnzhyee opened this issue · 1 comments

wnzhyee commented

Thanks for your great contribution！@taokong
I have some issues when i train and test RON with other dataset:
(1)How to use multi gpus while im training and testing?i find that in 'train_net.py' or 'test_net.py' ,the GPUS argument can only get one int32 value.
(2)If i only use 1 gpu when im testing ,it seems the caffe always out of the memory with the ERROR:status == CUDNN_STATUS_SUCCESS (8 vs 0)CUDNN_STATUS_EXECUTION_FAILED , so where can i reduce the batchsize when im testing ?
Thank you !

taokong commented

Sorry, this version of RON does not support multiple gpu training. However, you could add this feature with NCCL, I think https://github.com/bharatsingh430/py-R-FCN-multiGPU/ will help.

For not using gpu 0 for testing, you should also change the lib/fast_rcnn/config.py in line 117 to your gpu id.