Research map / Detection and segmentation / Tracking
MeMOT: Multi-Object Tracking With Memory
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 |
|---|---|---|---|
| Use MeMOT's memory-based joint detection and association | Online multi-object tracking · MOT16 | MOTA: 69.3 -> 72.6 | Helped |
| Use MeMOT's memory-based joint detection and association | Online multi-object tracking · MOT17 | IDF1: 73.6 -> 69.0 | Hurt |
| Use MeMOT's memory-based joint detection and association | Online multi-object tracking · MOT20 | IDF1: 67.3 -> 66.1 | Hurt |
| Use short-term memory length T_s=2 rather than T_s=3 (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 73.15 -> 72.52 | Hurt |
| Use short-term memory length T_s=5 rather than T_s=3 (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 73.15 -> 72.75 | No clear effect |
| Increase long-term memory length from T_l=3 to T_l=24 (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 71.27 -> 73.15 | Helped |
| Replace attention aggregation with average pooling over T=3 (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 73.15 -> 25.04 | Hurt |
| Replace separate short- and long-term branches with one cross-attention module (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 73.15 -> 72.64 | Hurt |
| Use short-term aggregate to query long-term memory (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 73.15 -> 70.30 | Hurt |
| Add dynamic updates to memory aggregation tokens (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: 61.03 -> 73.15 | Helped |
| Add a separate uniqueness score alongside objectness (ablation) | MeMOT decoder confidence prediction · MOT15 training validation | IDF1: 69.09 -> 73.15 | Helped |
| Use the latest observation as the short-term query and learnable tokens as the long-term query (ablation) | MeMOT memory aggregator · MOT15 training validation | IDF1: +0.85 IDF1 | Helped |
Closest papers by meaning
- A Distractor-Aware Memory for Visual Object Tracking with SAM2 (CVPR 2025 · similarity 0.88)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.88)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.87)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.87)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.87)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.86)
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.