Boxless Action Recognition in Still Images via Recurrent Visual Attention.
Weijiang FengXiang ZhangXuhui HuangZhigang LuoPublished in: ICONIP (2) (2017)
Keyphrases
- feature vectors
- action recognition
- image features
- visual attention
- static images
- salient regions
- saliency map
- feature space
- image data
- input image
- bag of words
- image retrieval
- eye tracking
- eye movements
- natural scenes
- image classification
- higher level
- activity recognition
- body parts
- computer vision
- human detection
- object recognition
- three dimensional
- human actions
- vision system
- image collections
- depth information
- low level