We have analyzed nearly 700,000 faces obtained from social media posts, this repository contains face attributes(age, gender, race, and etc.) and facial landmarks for each face. We hope this face library could be helpful for any research involving human face.
Hi
Thanks for sharing the code.
But I cann't view the image through the browser with the example
"https://scontent-ort2-1.cdninstagram.com/t51.2885 15/s640x640/sh0.08/e35/17077495_255710568218080_4361264754778439680_n.jpg".
I converted the image IDs into image URLs, and I've noticed that many of the images are invalid. For example 3579832224_1364722965090736174 from face_lib_16.txt becomes https://instagram.com/p/BLweE5fDRgu
Currently I see 91903 out of 315966 images are 404, which is around 29% of the images. Does this sound right, or am I making a mistake?
I also noticed that 56 lines in the source files are newlines, and they will cause a JSON parser to throw an error when attempting to decode them
hi, xuefeng:
thanks for sharing so many face data. But i can't find the face position in according to the origin pictue. Any hint to me will be appericiated.