Caption-Supervised Face Recognition: Training a State-of-the-Art Face Model Without Manual Annotation.
Qingqiu HuangLei YangHuaiyi HuangTong WuDahua LinPublished in: ECCV (17) (2020)
Keyphrases
- active shape model
- face model
- manual annotation
- facial features
- face recognition
- face images
- human faces
- supervised learning
- shape model
- fully automatic
- facial expressions
- automatic analysis
- automatic annotation
- face detection
- facial images
- training set
- semi automatic
- active appearance models
- manually annotated
- error prone
- computer vision
- automatic image annotation
- learning algorithm
- feature selection
- face databases
- image annotation
- lighting conditions
- data sets
- visual features
- feature points
- semi supervised
- training data
- pose variations
- relation extraction
- feature space
- principal component analysis