人脸属性指的是根据给定的人脸判断其性别、年龄和表情等,当前在github上开源了一些相关的工作,大部分都是基于tensorflow的,还有一部分是keras,CVPR2015曾有一篇是用caffe做的。
1、CVPR2015 caffe实现
https://github.com/GilLevi/AgeGenderDeepLearning
2、CVPR2015对应的tensorflow实现
https://github.com/dpressel/rude-carnie
3、DEX: Deep EXpectation 实现
https://github.com/truongnmt/multi-task-learning
4、CVPR2017 Age progression/regression by conditional Adversarial Autoencoder
https://github.com/ZZUTK/Face-Aging-CAAE
5、使用inception v1同时预测性别和年龄,受限于使用的dlib检测器,效果并不是很好
https://github.com/BoyuanJiang/Age-Gender-Estimate-TF
6、性别种族识别:gender Accuracy: 0.951493,race Accuracy: 0.87557212
https://github.com/zZyan/race_gender_recognition
8、性别识别全流程实现:94% accuracy
https://github.com/jocialiang/gender_classifier
9、表情、性别识别(keras)
https://github.com/oarriaga/face_classification
注明:两个分开的模型对应表情识别和性别识别
10、性别和种族识别
https://github.com/wondonghyeon/face-classification
11、年龄识别
https://github.com/shamangary/SSR-Net
12、年龄识别(亚洲人优化)
https://github.com/b02901145/SSR-Net_megaage-asian
13、年龄和性别识别(Keras )
https://github.com/yu4u/age-gender-estimation
14、表情和性别识别:表情66% with fer2013
https://github.com/isseu/emotion-recognition-neural-networks
15、性别和年龄识别(tensorflow ):91% accuracy in gender and 55% in age
https://github.com/zealerww/gender_age_classification
16、人脸检测、性别和表情识别、数字化妆、轮廓标识等多功能(tensorflow 、keras)
https://github.com/vipstone/faceai
17、表情识别(tensorflow、keras):fer2013
https://github.com/XiuweiHe/EmotionClassifier
18、表情和种族识别:表情 72% accuracy ,种族95% accuracy
https://github.com/HectorAnadon/Face-expression-and-ethnic-recognition
19、表情识别(caffe):66.7% on fer2013 with resnet50
https://github.com/ybch14/Facial-Expression-Recognition-ResNet
20、表情识别(keras)
https://github.com/JostineHo/mememoji
21、种族识别
https://github.com/mangorocoro/racedetector
22、多任务学习:性别、年龄、表情(tensorflow)
https://github.com/truongnmt/multi-task-learning
23、性别识别(tensorflow、keras):爬虫图片+人脸提取识别
https://github.com/StevenKe8080/recognition_gender
24、年龄识别(tensorflow)
https://github.com/zonetrooper32/AgeEstimateAdience
25、年龄和性别识别
https://github.com/OValery16/gender-age-classification