Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Multi-Shot Pedestrian Re-Identification via Sequential Decision Making
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use sequential RL decisions with an unsure action rather than average-pooling every frame | Inception-BN pedestrian re-identification · PRID2011 | CMC Rank 1: 84.3 -> 85.2 | Helped |
| Use sequential RL decisions with an unsure action rather than average-pooling every frame | Inception-BN pedestrian re-identification · iLIDS-VID | CMC Rank 1: 60.0 -> 60.2 | No clear effect |
| Use sequential RL decisions with an unsure action rather than average-pooling every frame | Inception-BN pedestrian re-identification · MARS | CMC Rank 1: 68.3 -> 71.2 | Helped |
| Set the unsure-action reward to 0.2 (ablation) | RL aggregation with Inception-BN · iLIDS-VID | CMC Rank 1: 60.3 (r_p=0.1) -> 60.2 (r_p=0.2) | No clear effect |
| Set the unsure-action reward to 0.2 (ablation) | RL aggregation with Inception-BN · PRID2011 | CMC Rank 1: 80.8 -> 85.2 | Helped |
| Set the unsure-action reward to 0.2 (ablation) | RL aggregation with Inception-BN · MARS | CMC Rank 1: 67.5 -> 71.2 | Helped |
| Add the three handcrafted distance features (ablation) | RL aggregation · PRID2011 | CMC Rank 1: 83.5 -> 85.2 | Helped |
| Add the three handcrafted distance features (ablation) | RL aggregation · iLIDS-VID | CMC Rank 1: 57.8 -> 60.2 | Helped |
| Add the three handcrafted distance features (ablation) | RL aggregation · MARS | CMC Rank 1: 69.2 -> 71.2 | Helped |
| Replace the fully connected layer with an LSTM layer for history aggregation (ablation) | RL aggregation · PRID2011 | CMC Rank 1: 85.2 -> 83.2 | Hurt |
| Feed image pairs sequentially instead of in random order (ablation) | RL aggregation · PRID2011 | CMC Rank 1: 85.2 -> 84.1 | Hurt |
| Fine-tune with average-pooled features from 8 sampled frames (ablation) | Inception-BN image extractor · MARS | CMC Rank 1: 71.2 -> 70.7 | Hurt |
Closest papers by meaning
- Multi-Rate Gated Recurrent Convolutional Networks for Video-Based Pedestrian Re-Identification (AAAI 2018 · similarity 0.90)
- Temporal Complementarity-Guided Reinforcement Learning for Image-to-Video Person Re-Identification (CVPR 2022 · similarity 0.87)
- Exploit the Unknown Gradually: One-Shot Video-Based Person Re-Identification by Stepwise Learning (CVPR 2018 · similarity 0.85)
- Event-Guided Person Re-Identification via Sparse-Dense Complementary Learning (CVPR 2023 · similarity 0.85)
- Recurrent Convolutional Network for Video-Based Person Re-Identification (CVPR 2016 · similarity 0.85)
- Active Vision for Early Recognition of Human Actions (CVPR 2020 · similarity 0.84)
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.