search

gjy3035 / GCC-SFCN

This is the official code of spatial FCN in the paper Learning from Synthetic Data for Crowd Counting in the Wild [CVPR2019].
https://gjy3035.github.io/GCC-CL/
MIT License
160 stars 39 forks source link

RuntimeError: CUDA error: out of memory #10

Open 2829841249 opened 5 years ago

2829841249 commented 5 years ago

Hello, I was training because the computer has only one gpu0 reason to fail, and I have been on a tight schedule recently, and I want to run the test directly. Can you email me the model(all_ep_178_mae_74.1_mse_131.0.pth) you need for the test?The model is the line 36 in test.py.Thanks!

2829841249 commented 5 years ago

my email is 2829841249@qq.com

gjy3035 commented 5 years ago

This test code is part of our other unpublished paper in fact. I think our test code is only for visualization, so I post it. The model is on ShangHai Part A.

I have updated the test.py for resSFCN and uploaded the trained model on QNRF, namely CVPR version and the result of the screenshot in this repo. Please download them from this link. Thanks for your attention!

FanShuixing commented 5 years ago

@gjy3035 Hello,Thanks for your sharing. I met a problem when I want to run python train.py,the wrong information is: RuntimeError: CUDA out of memory. Tried to allocate 81.00 MiB (GPU 0; 15.75 GiB total capacity; 14.50 GiB already allocated; 39.94 MiB free; 461.00 KiB cached). I was using one Tesla V100 with 16memery ,and I want to know how much memory does the model need? thanks.

gjy3035 commented 5 years ago

@FanShuixing You would adopt our provided data for training, which is scaled down keep the original ratio. Also, you may need a smaller batch size for single GPU.