test my own image

[PMPBinZhang]

hello, how can i use my own image to test, and how can i get the tsv file? looking forward to your reply, thanks

[orizv]

Interested in the answer as well

[ZhenglinZhou]

Sorry for the late reply. The STAR loss was built on the top of ADNet, which is a strong baseline. For the own image, there are there important paras in tsv file, which are scale, center_x and center_y. In the discussion with the author of ADNet (really kind!), the key is to get the bounding box of face:

• scale = min(x2 – x1, y2 – y1) / 200 * 1.05
• center_x = (x1 + x2) / 2
• center_y = (y1 + y2) / 2