MMLA: Multi-Environment, Multi-Species, Low-Altitude Drone Dataset
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915568541499392 |
|---|---|
| author | Kline, Jenna Stevens, Samuel Maalouf, Guy Saint-Jean, Camille Rondeau Ngoc, Dat Nguyen Mirmehdi, Majid Guerin, David Burghardt, Tilo Pastucha, Elzbieta Costelloe, Blair Watson, Matthew Richardson, Thomas Lundquist, Ulrik Pagh Schultz |
| author_facet | Kline, Jenna Stevens, Samuel Maalouf, Guy Saint-Jean, Camille Rondeau Ngoc, Dat Nguyen Mirmehdi, Majid Guerin, David Burghardt, Tilo Pastucha, Elzbieta Costelloe, Blair Watson, Matthew Richardson, Thomas Lundquist, Ulrik Pagh Schultz |
| contents | Real-time wildlife detection in drone imagery supports critical ecological and conservation monitoring. However, standard detection models like YOLO often fail to generalize across locations and struggle with rare species, limiting their use in automated drone deployments. We present MMLA, a novel multi-environment, multi-species, low-altitude drone dataset collected across three sites (Ol Pejeta Conservancy and Mpala Research Centre in Kenya, and The Wilds in Ohio), featuring six species (zebras, giraffes, onagers, and African wild dogs). The dataset contains 811K annotations from 37 high-resolution videos. Baseline YOLO models show performance disparities across locations while fine-tuning YOLOv11m on MMLA improves mAP50 to 82%, a 52-point gain over baseline. Our results underscore the need for diverse training data to enable robust animal detection in autonomous drone systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_07744 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MMLA: Multi-Environment, Multi-Species, Low-Altitude Drone Dataset Kline, Jenna Stevens, Samuel Maalouf, Guy Saint-Jean, Camille Rondeau Ngoc, Dat Nguyen Mirmehdi, Majid Guerin, David Burghardt, Tilo Pastucha, Elzbieta Costelloe, Blair Watson, Matthew Richardson, Thomas Lundquist, Ulrik Pagh Schultz Computer Vision and Pattern Recognition Real-time wildlife detection in drone imagery supports critical ecological and conservation monitoring. However, standard detection models like YOLO often fail to generalize across locations and struggle with rare species, limiting their use in automated drone deployments. We present MMLA, a novel multi-environment, multi-species, low-altitude drone dataset collected across three sites (Ol Pejeta Conservancy and Mpala Research Centre in Kenya, and The Wilds in Ohio), featuring six species (zebras, giraffes, onagers, and African wild dogs). The dataset contains 811K annotations from 37 high-resolution videos. Baseline YOLO models show performance disparities across locations while fine-tuning YOLOv11m on MMLA improves mAP50 to 82%, a 52-point gain over baseline. Our results underscore the need for diverse training data to enable robust animal detection in autonomous drone systems. |
| title | MMLA: Multi-Environment, Multi-Species, Low-Altitude Drone Dataset |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.07744 |