chenxi116 / DeepLabv3.pytorch

PyTorch implementation of DeepLabv3
Other
253 stars 47 forks source link

cityscapes result #6

Open erichhhhho opened 5 years ago

erichhhhho commented 5 years ago

Hi, I saw the segmentation results are mainly in Pascal VOC, anyone tried it on Cityscapes?

chenxi116 commented 5 years ago

Hi, you can try python main.py --train --dataset cityscapes --exp lr1e-2_crop769_batch20_lastmult10 --epochs 480 --base_lr 0.01 --crop_size 769 --batch_size 20 --last_mult 10, which should get >76 mIOU, which is ~1 lower than what is reported in the paper, unfortunately. I was using 4 GPUs, so batch size 5 on each GPU.

erichhhhho commented 5 years ago

Hi, you can try python main.py --train --dataset cityscapes --exp lr1e-2_crop769_batch20_lastmult10 --epochs 480 --base_lr 0.01 --crop_size 769 --batch_size 20 --last_mult 10, which should get >76 mIOU, which is ~1 lower than what is reported in the paper, unfortunately. I was using 4 GPUs, so batch size 5 on each GPU.

Hi, thx for your reply. How about the experiment with WS in Cityscapes?

chenxi116 commented 5 years ago

That I have not tried.