tianzhi0549 / FCOS

FCOS: Fully Convolutional One-Stage Object Detection (ICCV'19)
https://arxiv.org/abs/1904.01355
Other
3.28k stars 630 forks source link

Did you freeze parts of the backbone when training? #99

Closed ZhengMengbin closed 5 years ago

ZhengMengbin commented 5 years ago

I see MODEL.BACKBONE.FREEZE_CONV_BODY_AT = 2 in the defaults.py, so did you freeze parts of the backbone when training? image

tianzhi0549 commented 5 years ago

@ZhengMengbin Yes. Some layers are frozen.

ZhengMengbin commented 5 years ago

Thanks for your reply!