Research map / Detection and segmentation / Tracking
Detection and Tracking Meet Drones Challenge
Pengfei Zhu, Longyin Wen, Dawei Du, Xiao Bian, Heng Fan, Qinghua Hu, Haibin Ling
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 |
|---|---|---|---|
| use video object detectors with temporal information instead of image detectors | D&T vs FPN · VisDrone-VID test-dev | AP: 12.93 -> 14.21 | Improved |
| combine several effective modules on the Cascade R-CNN baseline | Cascade R-CNN · VisDrone-DET test-challenge 2020 | AP: 16.09 -> 34.57 | Improved |
| integrate deformable convolution and global context into Cascade R-CNN | Cascade R-CNN · VisDrone-VID test-challenge 2019 | AP: 16.09 -> 29.22 | Improved |
| train on large-scale datasets offline for trackers | ATOM · VisDrone-SOT test-dev | success score: 64.5 | Improved |
| shift from correlation filter to deep network trackers | VisDrone-SOT winners · VisDrone-SOT test-challenge 2018 | success score: 68.0 -> 75.5 | Improved |
| ensemble several trackers (ATOM, SiamRPN++, Siam R-CNN, DiMP) | LTNMI · VisDrone-SOT test-challenge 2018 | success score: 75.5 -> 76.5 | Improved |
| add squeeze-and-excitation to ATOM | ATOM · VisDrone-SOT test-challenge 2018 | success score: 75.5 | Improved |
| use newly collected long-term sequences (2019 set) | best SOT trackers · VisDrone-SOT | success score: 68.0 -> 48.9 | Hurt |
| improve trackers from 2019 to 2020 challenge | best SOT tracker · VisDrone-SOT test-challenge 2019 | success score: 48.9 -> 55.5 | Improved |
| use video detector over detector improvement via feature enhancement (SSD) | CFE-SSDv2 · VisDrone-VID test-challenge 2018 | AP: 8.63 -> 21.57 | Improved |
| add tracking (ECO) to associate detections | SSD · VisDrone-VID test-challenge 2018 | AP: 21.57 -> 16.54 | Mixed |
| remove higher convolutional layers and add attention to RetinaNet | RetinaNet · VisDrone-DET test-challenge 2018 | AP: 18.94 -> 31.88 | Improved |
Closest papers by meaning
- Visual Object Tracking for Unmanned Aerial Vehicles: A Benchmark and New Motion Models (AAAI 2017 · similarity 0.90)
- UAV3D: A Large-scale 3D Perception Benchmark for Unmanned Aerial Vehicles (NeurIPS 2024 · similarity 0.88)
- MMOT: The First Challenging Benchmark for Drone-based Multispectral Multi-Object Tracking (NeurIPS 2025 · similarity 0.87)
- MUST: The First Dataset and Unified Framework for Multispectral UAV Single Object Tracking (CVPR 2025 · similarity 0.86)
- PointSR: Self-Regularized Point Supervision for Drone-View Object Detection (CVPR 2025 · similarity 0.85)
- Multiview Aerial Visual RECognition (MAVREC): Can Multi-view Improve Aerial Visual Perception? (CVPR 2024 · similarity 0.85)
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.
