Multi-stream Convolutional Neural Networks for Action Recognition in Video Sequences Based on Adaptive Visual Rhythms.
Darwin Ttito ConchaHelena de Almeida MaiaHélio PedriniHemerson TaconAndré de Souza BritoHugo de Lima ChavesMarcelo Bernardes VieiraPublished in: ICMLA (2018)
Keyphrases
- action recognition
- human actions
- multi stream
- convolutional neural networks
- video sequences
- body parts
- visual data
- mid level
- audio visual
- bag of words
- space time interest points
- activity recognition
- video dataset
- computer vision
- visual information
- audio visual speech recognition
- visual features
- action classification
- recognizing human actions
- bag of features
- motion estimation
- image sequences
- space time
- view invariant
- low level
- recognition of human actions
- action recognition in videos
- motion capture data
- moving objects
- human pose
- action detection
- recognizing actions
- object tracking
- video frames
- feature extraction