Research map / Video / Video segmentation and tracking
Temporal Complementarity-Guided Reinforcement Learning for Image-to-Video Person Re-Identification
Check 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 |
|---|---|---|---|
| Add TCRL with complementary feature extraction and adaptive frame selection | Image-to-video person Re-ID · MARS | Rank-1: 83.9 -> 86.0 | Helped |
| Add TCRL with complementary feature extraction and adaptive frame selection | Image-to-video person Re-ID · iLIDS-VID | Rank-1: 54.7 -> 77.3 | Helped |
| Add sequential feature extraction with complementary residual detectors (ablation) | ResNet-50 image-to-video Re-ID on MARS · MARS | Rank-1: 82.8 -> 85.3; 2.5% | Helped |
| Add sequential feature extraction with complementary residual detectors (ablation) | ResNet-50 image-to-video Re-ID on MARS · MARS | mAP: 74.4 -> 78.8; 4.4% | Helped |
| Add the sequential judgement agent for adaptive frame selection (ablation) | Basel+SFE on MARS · MARS | Rank-1: 85.3 -> 86.0; 0.7% | Helped |
| Add the sequential judgement agent for adaptive frame selection (ablation) | Basel+SFE on MARS · MARS | mAP: 78.8 -> 80.1; 1.3% | Helped |
| Set the agent reward parameter β to 1.8 instead of 1.9 (ablation) | TCRL sequential judgement agent · MARS | Rank-1: 86.0 -> 84.3 | Hurt |
| Set the agent reward parameter β to 2.0 instead of 1.9 (ablation) | TCRL sequential judgement agent · MARS | Rank-1: 86.0 -> 85.2 | Hurt |
| Set the input clip length T to 5 instead of 6 (ablation) | TCRL sequential feature extraction · MARS | Rank-1: 85.3 -> 84.3 | Hurt |
| Set the input clip length T to 7 instead of 6 (ablation) | TCRL sequential feature extraction · MARS | mAP: 78.8 -> 77.6 | Hurt |
| Set salient convolutional kernel size k to 1 instead of 3 (ablation) | TCRL sequential feature extraction · MARS | Rank-1: 85.3 -> 84.4 | Hurt |
| Set salient convolutional kernel size k to 5 instead of 3 (ablation) | TCRL sequential feature extraction · MARS | mAP: 78.8 -> 78.2 | Hurt |
Closest papers by meaning
- Multi-Shot Pedestrian Re-Identification via Sequential Decision Making (CVPR 2018 · similarity 0.87)
- AXM-Net: Implicit Cross-Modal Feature Alignment for Person Re-identification (AAAI 2022 · similarity 0.86)
- Deep Neural Networks with Inexact Matching for Person Re-Identification (NeurIPS 2016 · similarity 0.86)
- Learning What to Generate: A Reinforcement Learning-based Closed-Loop Augmentation Framework for Person Re-identification (ICML 2026 · similarity 0.86)
- Temporal Correlation Vision Transformer for Video Person Re-Identification (AAAI 2024 · similarity 0.84)
- TF-CLIP: Learning Text-Free CLIP for Video-Based Person Re-identification (AAAI 2024 · 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.