search

tianzhi0549 / FCOS

FCOS: Fully Convolutional One-Stage Object Detection (ICCV'19)
https://arxiv.org/abs/1904.01355
Other
3.27k stars 630 forks source link

centerness shared head #89

Open AllenPeng0209 opened 5 years ago

AllenPeng0209 commented 5 years ago

Hi,

I want to know that is there any experiment results show that centerness branch shared the same parameter with classification branch achieve the best MAP?

Would it be better to separate the centerness branch?

And what it the effect and result behind?

Thanks !

tianzhi0549 commented 5 years ago

@AllenPeng0209 It is discussed in Section 4.1.2 in our paper.

AllenPeng0209 commented 5 years ago

Also,

would it be better to train 80 centerness scores for 80 class instead of 1 score for 80 class? Thanks

tianzhi0549 commented 5 years ago

@AllenPeng0209 Sorry, we have not tried this. You might try it by yourself.

bearcatt commented 5 years ago

@tianzhi0549 I've tried sharing the center-ness head with the reg head rather than the cls head. This simple modification leads to 0.5 mAP gain for FCOS-R101. You can try this.

tianzhi0549 commented 5 years ago

@bearcatt Thank you very much! We will try it.

tianzhi0549 commented 5 years ago

@bearcatt We have confirmed it can improve the performance. Do you mind we add this into our code with a proper acknowledgement?

bearcatt commented 5 years ago

@tianzhi0549 that's great! You can also refer to the code here.

tianzhi0549 commented 5 years ago

@bearcatt Thank you:-). Let me do it.

18804601171 commented 4 years ago

@bearcatt Why I share center-ness head in reg branch at mmdetection,it's not improved

JunsukLee commented 2 years ago

The value of CENTERNESS_ON_REG in the provided yaml file is set to True. The performance of the paper shows excellent performance when CENTERNESS_ON_REG:False. If the default value of all yaml is True, is the performance excellent? I have confused.