get poor performance of ConInst

aim-uofa / AdelaiDet

AdelaiDet is an open source toolbox for multiple instance-level detection and recognition tasks.

https://git.io/AdelaiDet

Other

3.37k stars 646 forks source link

get poor performance of ConInst #219

Closed jiangzhengkai closed 3 years ago

jiangzhengkai commented 3 years ago

Hi,

I just run the default ConInst with condinst_MS_R_50_1x config, 8 GPUs, and 16 batch sizes in total as described in this repo. However, I get poor performance.

Here is log log. Can you know where is the problem?

tianzhi0549 commented 3 years ago

@jiangzhengkai The log file requires permission to access.

jiangzhengkai commented 3 years ago

@tianzhi0549 Sorry, How about now?

tianzhi0549 commented 3 years ago

@jiangzhengkai It seems you only used one GPU according to your log. Could you please post your training command line here?

jiangzhengkai commented 3 years ago

python3 tools/train_net.py --config-file configs/CondInst/MS_R_50_1x.yaml You are right, I forget to add --num-gpus parameters in command. Thanks for your quick reply!