OAS-Net: Occlusion Aware Sampling Network for Accurate Optical Flow.
Lingtong KongXiaohang YangJie YangPublished in: ICASSP (2021)
Keyphrases
- optical flow
- peer to peer
- network model
- high quality
- computationally efficient
- monte carlo
- complex networks
- network traffic
- motion field
- network architecture
- dense optical flow
- occlusion boundaries
- optical flow computation
- network management
- computer networks
- motion boundaries
- sampling strategy
- flow field
- three dimensional
- network structure
- image sequences