microsoft/MegaDetector-Overhead

microsoft/MegaDetector-Overhead

Description: MegaDetector-Overhead — The Microsoft open-source AI for overhead wildlife detection. Point-based detection model for aerial and drone imagery, identifying wildlife from above. Maintained by Microsoft AI for Good Lab. Part of the Pytorch-Wildlife ecosystem.

Language: Python

License: MIT

Stars: 1

Forks: 0

Open issues: 6

Created: 2026-05-06T02:36:58Z

Pushed: 2026-06-05T19:30:28Z

Default branch: main

Fork: no

Archived: no

README:

MegaDetector-Overhead

Open-source AI for detecting wildlife in overhead and aerial imagery.

MegaDetector-Overhead extends the MegaDetector detection framework to drone and UAV survey imagery, handling the unique challenges of overhead perspectives: small targets, variable altitude, and nadir-angle distortion. It is powered by PyTorch-Wildlife and is part of the microsoft/Biodiversity ecosystem.

This repository ships the training, evaluation, and inference stack for the OWL model family:

| Model | Backbone | Notes | |---|---|---| | OWL-C | DLA-34 (HerdNet detection branch) | Baseline; fast inference | | OWL-T | DLA-34 + Swin transformer multiscale residual | Sharper localization on cluttered backgrounds | | OWL-D (S / B / L / H) | DINOv3 ViT + DPT decoder | Highest quality; foundation-model encoder |

The legacy HerdNet multi-class model is also available. See [Model Zoo](docs/model_zoo.md) for the full list.

---

Documentation

Full documentation at [microsoft.github.io/MegaDetector-Overhead](https://microsoft.github.io/MegaDetector-Overhead/)

• [Installation](INSTALL.md) — full install + DINOv3 weights download
• [Training, Evaluation, and Inference](docs/training.md) — end-to-end workflow

---

Quick Start

The environment is managed with uv. One uv sync builds a Python 3.11 venv with all dependencies, the animaloc training package, and the vendored DINOv3 encoder.

# 1. Install uv (one-time)
curl -LsSf https://astral.sh/uv/install.sh | sh

# 2. Clone and sync
git clone https://github.com/microsoft/MegaDetector-Overhead
cd MegaDetector-Overhead
uv sync

# 3. Smoke test
uv run python -c "import animaloc.models, dinov3; print('OK')"

See [INSTALL.md](INSTALL.md) for DINOv3 weights download and troubleshooting.

---

Repository Layout

animaloc/ # Training/eval package vendored from HerdNet (MIT)
dinov3/ # DINOv3 encoder vendored from facebookresearch/dinov3 (DINOv3 License)
tools/ # train.py, test.py, infer.py, patcher.py
configs/ # Hydra configs for OWL-C / OWL-D / OWL-T training and eval
docs/ # MkDocs Material site (build with `uv run --extra docs mkdocs build`)

See [NOTICE](NOTICE) for upstream attribution and third-party licenses.

---

Ecosystem

| Repository | Description | |---|---| | microsoft/Biodiversity | Umbrella hub — PyTorch-Wildlife, MegaDetector, ecosystem overview | | microsoft/MegaDetector | Animal, human, and vehicle detection for camera-trap images | | microsoft/PytorchWildlife | The collaborative deep learning framework for wildlife monitoring | | microsoft/MegaDetector-Overhead | This repo — wildlife detection in aerial and drone imagery | | microsoft/MegaDetector-Acoustic | Bioacoustic AI for audio-based wildlife monitoring | | microsoft/MegaDetector-Sonar | Sonar-based wildlife detection for aquatic monitoring | | microsoft/SPARROW | Solar-Powered Acoustic and Remote Recording Observation Watch |

---

Citation

If you use MegaDetector-Overhead in your research, please cite the framework, the vendored HerdNet training stack, and the DINOv3 backbone. Full BibTeX in [docs/cite.md](docs/cite.md).

@misc{hernandez2024pytorchwildlife,
title={Pytorch-Wildlife: A Collaborative Deep Learning Framework for Conservation},
author={Andres Hernandez and Zhongqi Miao and Luisa Vargas and Sara Beery and Rahul Dodhia and Juan Lavista},
year={2024},
eprint={2405.12930},
archivePrefix={arXiv},
}

@article{delplanque2023herdnet,
title = {From crowd to herd counting: How to precisely detect and count African mammals using aerial imagery and deep learning?},
journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
volume = {197},
pages = {167-180},
year = {2023},
doi = {10.1016/j.isprsjprs.2023.01.025},
author = {Alexandre Delplanque and Samuel Foucher and Jérôme Théau and Elsa Bussière and Cédric Vermeulen and Philippe Lejeune}
}