Unsupervised Neural Quantization for Compressed-Domain Similarity Search.
Stanislav MorozovArtem BabenkoPublished in: ICCV (2019)
Keyphrases
- similarity search
- compressed domain
- pixel domain
- motion vectors
- video analysis
- metric space
- dct domain
- high dimensional
- query processing
- bitstream
- similarity measure
- neural network
- similarity searching
- inter frame
- dct coefficients
- similarity queries
- knn
- jpeg images
- high dimensional data
- semi supervised
- motion estimation
- approximate similarity search
- coding scheme
- bit rate
- cross view
- video coding
- metric access methods
- hash functions
- low dimensional
- feature vectors
- computational complexity
- training data
- e learning