search

ethnhe / PVN3D

Code for "PVN3D: A Deep Point-wise 3D Keypoints Hough Voting Network for 6DoF Pose Estimation", CVPR 2020
MIT License
488 stars 105 forks source link

Adjusting Batchsize ? #51

Closed BenjaminBo closed 3 years ago

BenjaminBo commented 3 years ago

When running python3 -m train.train_linemod_pvn3d --cls ape I get the following error: RuntimeError: CUDA out of memory. Tried to allocate 1.76 GiB (GPU 0; 23.62 GiB total capacity; 19.67 GiB already allocated; 815.44 MiB free; 1.94 MiB cached)

The only other process running on the GPU is a python process which takes up 2317 MiB out of 23.62 GiB. There has to be a way to run PVN3D without taking up 20 GiB, right? Maybe by specifying a BATCH_SIZE somewhere? I'm new to the CUDA world so I'm not sure how or where I would have to adjust/specify that. Could somebody help me with that?

BenjaminBo commented 3 years ago

batchsizes have to be adjusted in common.py.