endernewton / tf-faster-rcnn

Tensorflow Faster RCNN for Object Detection
https://arxiv.org/pdf/1702.02138.pdf
MIT License
3.65k stars 1.58k forks source link

bounding box regression #475

Open ghost opened 4 years ago

ghost commented 4 years ago

tx = (x -xa) / wa , ty = (y-ya) / ha, tw = log(w/wa) , th = log(h/ha)

1 . In the trainning processing , the updated variables are x , y , w , h ,right ? then tx is closing to tx* . 2 . the fixed variable is xa,ya,wa,ha . because the anchor is fixed which not influenced by the trainning processing. is right ?