search

liuzhengzhe / 3D-to-2D-Distillation-for-Indoor-Scene-Parsing

CVPR 2021 Oral https://arxiv.org/abs/2104.02243
47 stars 4 forks source link

validating efficiency #7

Open lqzhao opened 3 years ago

lqzhao commented 3 years ago

Hi, thanks for your great work. I found the validating efficiency is not good because it will cost about 10 hours to test each image with 6 scales. I reduce the number of scales to 1, i.e., 1.0, but it still costs about 3 hours. So here is the question:

Can we just set a larger batch size when testing? The current version doesn't support a larger batch size than 1.

liuzhengzhe commented 3 years ago

Good idea. I think it is feasible.

lqzhao @.***> 于2021年8月23日周一 下午11:26写道：

Hi, thanks for your great work. I found the validating efficiency is not good because it will cost about 10 hours to test each image with 5 scales. I reduce the number of scales to 1, i.e., 1.0, but it still costs about 3 hours. So here is the question:

Can we just set a larger batch size when testing? The current version doesn't support a larger batch size than 1.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/liuzhengzhe/3D-to-2D-Distillation-for-Indoor-Scene-Parsing/issues/7, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB7WPDDQ5M7RL75OMB4EO7LT6JSD7ANCNFSM5CU2NSCA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email .