Human Action Recognition Using Optical Flow and Convolutional Neural Networks.
Marco WrzalikDirk KrechelPublished in: ICMLA (2017)
Keyphrases
- convolutional neural networks
- optical flow
- convolutional network
- image sequences
- optical flow estimation
- flow field
- motion estimation
- feature tracking
- optical flow computation
- motion model
- least squares
- dense optical flow
- d scene
- moving objects
- qualitative and quantitative evaluation
- motion field
- computer vision
- image motion
- velocity field
- motion parameters
- motion segmentation
- vector field
- data sets
- ego motion
- estimation of optical flow
- optical flow algorithm
- image segmentation
- motion discontinuities
- feature extraction
- optical flow fields
- structure tensor
- moving camera
- structure from motion
- camera motion
- video sequences