Research map / Detection and segmentation / Tracking
VETRA: A Dataset for Vehicle Tracking in Aerial Imagery - New Challenges for Multi-Object Tracking
Jens Hellekes, Manuel Mühlhaus, Reza Bahmanyar, Seyed Majid Azimi, Franz Kurz
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 |
|---|---|---|---|
| add BRISK camera motion compensation and adaptive search radius to Deep SORT (Deep SR-SORT) | Deep SORT · VETRA test set | HOTA: 58.29 -> 82.18 | Improved |
| add BRISK CMC and adaptive search radius to Deep SORT | Deep SORT · VETRA test set | IDSW: 3,077 -> 792 | Improved |
| include background (HBB) instead of OBB-masked crops for Re-ID (ablation) | ResNet50 Re-ID · VETRA test set | Re-ID mAP: 68.3 -> 73.2 | Improved |
| use Re-ID with OBB-masked background in Deep SORT (ablation) | Deep SORT · VETRA test set | HOTA: 58.29 -> 54.92 | Hurt |
| replace IoU with DIoU in Deep OC-SORT (ablation) | Deep OC-SORT · VETRA test set | HOTA: 46.84 -> 39.47 | Hurt |
| add Re-ID and camera motion compensation to motion-only tracker (BoT-SORT vs ByteTrack) | ByteTrack · VETRA test set | HOTA: 36.37 -> 50.84 | Improved |
| rely on Kalman motion models in Deep OC-SORT instead of Re-ID only (Deep SORT) | Deep SORT · VETRA test set | HOTA: 58.29 -> 46.84 | Hurt |
| train detector jointly on training and validation sets | DINO OBB detector · VETRA test set | mAP: 70.56 -> 72.62 | Improved |
| train on VETRA instead of VisDrone and test on VisDrone | Deep SR-SORT · VisDrone (GT boxes) | IDSW: 183 -> 374 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Byte-Track — ByteTrack: Multi-Object Tracking by Associating Every Detection Box (ECCV 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Visual Object Tracking for Unmanned Aerial Vehicles: A Benchmark and New Motion Models (AAAI 2017 · similarity 0.90)
- Detection and Tracking Meet Drones Challenge (TPAMI 2021 · similarity 0.87)
- MMOT: The First Challenging Benchmark for Drone-based Multispectral Multi-Object Tracking (NeurIPS 2025 · similarity 0.86)
- Learning Adaptive and View-Invariant Vision Transformer for Real-Time UAV Tracking (ICML 2024 · similarity 0.85)
- MUST: The First Dataset and Unified Framework for Multispectral UAV Single Object Tracking (CVPR 2025 · similarity 0.85)
- UAV3D: A Large-scale 3D Perception Benchmark for Unmanned Aerial Vehicles (NeurIPS 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.
