- [Paper Reading] Aggregating loca
- Bitcoin: A peer-to-peer electron
- Summary of the a paper ‘A love-
- 2019-11-04 Generating Sentences
- 论文阅读——Attention-based Transactio
- Multi-Person Pose Estimation wit
- Bag of Tricks for Image Classifi
- MixConv: Mixed Depthwise Convolu
- Uncertainty in multitask learnin
- Fast Human Pose Estimation
Paper Site: https://lear.inrialpes.fr/pubs/2010/JDSP10/jegou_compactimagerepresentation.pdf
Problem Definition
To jointly solve the three constraints: the accuracy of the search, the efficiency and the memory usage of the representation in the image search on a very large scale.
Contribution and Discussion
-
Propose a simple yet efficient way of aggregating local image descriptors into a vector of limited dimension, which can be viewed as a simplification of the Fisher kernel representation.
-
Jointly optimize the dimension reduction and the indexing algorithm, so that it best preserves the quality of vector comparison.
-
Significantly outperforms the state of the art: the search accuracy is comparable to the bag-of-features approach for an image representation that fits in 20 bytes. Searching a 10 million image dataset takes about 50ms.
Method
-
From vectors to codes: optimize 1) a projection that reduces the dimensionality of the vector and 2) a quantization used to index the resulting vectors.
-
Dimensionality reduction in approximate nearest neighbor search. Use principal component analysis (PCA) for dimensionality reduction.
-
Allocating different numbers of bits to the different components to balance the components' variance.
网友评论