1.2Faster RCNN源碼解析(pytorch)
2023-07-20 00:33 作者:卷坤神經(jīng)網(wǎng)絡(luò) | 我要投稿
根據(jù)彈幕老哥提示
anchor
scale * scale = hs * ws =
那么可以根據(jù) scale * scale = ws * ws * aspect_ratios 可得 ws = scale / 根號下 aspect_ratios 也就是說w_ratio= 1/根號下aspect_ratios
(scale / 根號下aspect_ratios )[也就是ws] * hs = scale *scale
可得 hs = scale * 根號下aspect_ratios
即h_ratios = 根號下aspect_ratios
標簽：
