tztztztztz / eqlv2

The official implementation of Equalization Loss v1 & v2 (CVPR 2020, 2021) based on MMDetection. https://arxiv.org/abs/2012.08548 https://arxiv.org/abs/2003.05176
Apache License 2.0
158 stars 23 forks source link

About baseline model #2

Open U-Help opened 3 years ago

U-Help commented 3 years ago

I have great interest in your work, and I want to ask a question. Does the baseline model also use sigmoid loss function rather than softmax loss function?

tztztztztz commented 3 years ago

Thanks for your interest. In this repo, the baseline model is trained with softmax loss function.

U-Help commented 3 years ago

Thanks for your reply. In your paper, at this location, you refer baseline model to sigmoid loss function, image whereas at corresponding picture, it becomes softmax loss function. I'm a little confused. image

tztztztztz commented 3 years ago

Thanks for pointing out this typo. In Figure 2, we use the baseline models with the sigmoid loss function.
We have already corrected it in the camera-ready version, which will be updated to arxiv soon.

ShirleyHe2020 commented 3 years ago

Hi Authors, can this work be combined into YOLO v5 ?