Table 1 Characteristics of recent deep learning hashing methods
ID | Method | Category | Network | Layer | Feature+hashing |
|---|---|---|---|---|---|
1 | Deep learning of binary hash [68] | Supervised | Single | 8 | Separate |
2 | Deep hashing for compact binary codes [69] | Unsupervised/supervised | Single | 3 | Separate |
3 | Unsupervised deep neural networks hashing [23] | Unsupervised | Single | 16 | Separate |
4 | Supervised deep hashing [22] | Supervised | Single | 5 | Together |
5 | Semantics-preserving deep hashing [23] | Supervised | Single | 8 | Together |
6 | Deep semantic ranking hashing [24] | Supervised | Single | 8 | Together |
7 | Binary deep neural network hashing [26] | Unsupervised/supervised | Single | 5 | Together |
8 | Bit-scalable deep hashing [27] | Triplet | Parallel | 10 | Together |
9 | One-stage deep hashing [28] | Triplet | Parallel | 10 | Together |
10 | Deep pairwise-supervised hashing [29] | Pairwise | Parallel | 8 | Together |
11 | Deep hashing network [72] | Pairwise | Parallel | 8 | Together |
12 | Deep semantic-preserving and ranking-based hashing [73] | Triplet | Parallel | 19 | Together |