Research map / Architectures and efficiency / RNNs and normalization
A Siamese Long Short-Term Memory Architecture for Human Re-identification
Rahul Rama Varior, Bing Shuai, Jiwen Lu, Dong Xu, Gang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace the feed-forward mapping with an LSTM over row features (LOMO) (ablation) | Siamese baseline 1 layer · Market-1501 | Rank-1: 46.9 -> 51.8 | Improved |
| replace the feed-forward mapping with an LSTM over row features (LOMO) (ablation) | Siamese baseline 1 layer · CUHK03 (detected) | Rank-1: 49.1 -> 55.8 | Improved |
| replace the feed-forward mapping with an LSTM over row features (LOMO) (ablation) | Siamese baseline 1 layer · VIPeR | Rank-1: 35.8 -> 40.5 | Improved |
| replace the feed-forward mapping with an LSTM over row features (LOMO+CN) (ablation) | Siamese baseline 1 layer · Market-1501 | Rank-1: 52.1 -> 61.6 | Improved |
| replace the feed-forward mapping with an LSTM over row features (LOMO+CN) (ablation) | Siamese baseline 1 layer · CUHK03 (detected) | Rank-1: 51.6 -> 57.3 | Improved |
| replace the feed-forward mapping with an LSTM over row features (LOMO+CN) (ablation) | Siamese baseline 1 layer · VIPeR | Rank-1: 36.1 -> 42.4 | Improved |
| add Color Names features to LOMO (ablation) | Siamese LSTM 1 layer · Market-1501 | Rank-1: 51.8 -> 61.6 | Improved |
| add Color Names features to LOMO (ablation) | Siamese LSTM 1 layer · CUHK03 (detected) | Rank-1: 55.8 -> 57.3 | Improved |
| add Color Names features to LOMO (ablation) | Siamese LSTM 1 layer · VIPeR | Rank-1: 40.5 -> 42.4 | Improved |
| increase baseline depth from 1 to 3 layers (LOMO) (ablation) | Siamese baseline · Market-1501 | Rank-1: 46.9 -> 48.4 | Improved |
| increase baseline depth from 1 to 3 layers (LOMO) (ablation) | Siamese baseline · VIPeR | Rank-1: 35.8 -> 34.8 | Hurt |
| increase baseline depth from 1 to 3 layers (LOMO) (ablation) | Siamese baseline · CUHK03 (detected) | Rank-1: 49.1 -> 51.1 | Improved |
Closest papers by meaning
- Gated Siamese Convolutional Neural Network Architecture for Human Re-identification (ECCV 2016 · similarity 0.91)
- Re-Identification With Consistent Attentive Siamese Networks (CVPR 2019 · similarity 0.90)
- Multi-Rate Gated Recurrent Convolutional Networks for Video-Based Pedestrian Re-Identification (AAAI 2018 · similarity 0.89)
- Dual Attention Matching Network for Context-Aware Feature Sequence Based Person Re-Identification (CVPR 2018 · similarity 0.89)
- Person Re-identification via Recurrent Feature Aggregation (ECCV 2016 · similarity 0.88)
- Recurrent Convolutional Network for Video-Based Person Re-Identification (CVPR 2016 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
