Unsupervised Learning of Optical Flow with Deep Feature Similarity.
Woobin ImTae-Kyun KimSung-Eui YoonPublished in: ECCV (24) (2020)
Keyphrases
- unsupervised learning
- optical flow
- deep learning
- deep architectures
- similarity measure
- image sequences
- supervised learning
- machine learning
- similarity measurement
- object recognition
- computer vision
- flow field
- euclidean distance
- neural network
- deep belief networks
- motion features
- image similarity
- feature tracking
- similarity metric
- similarity function
- distance measure
- dimensionality reduction
- motion estimation
- feature vectors
- motion field
- motion segmentation
- visual features
- feature set
- velocity field
- least squares
- optical flow computation
- video sequences
- multiscale