MegaDetector¶
MegaDetector is Microsoft's open-source AI model for camera-trap conservation — it detects animals, people, and vehicles in camera-trap imagery and filters out blank images, reducing manual review across large datasets. Developed and maintained by the Microsoft AI for Good Lab.
[!IMPORTANT] This page has moved. MegaDetector's canonical home is now microsoft/MegaDetector. Source code, releases, issues, the model zoo, the full version history (V1–V6), benchmarks, and the list of organizations using MegaDetector all live there. This page remains only as a pointer to the canonical repository.
Where things live¶
| You want… | Go to |
|---|---|
| MegaDetector source, releases, issues, V1–V6 history, benchmarks | microsoft/MegaDetector |
| The PyTorch-Wildlife framework that hosts MegaDetector and classifiers | microsoft/PytorchWildlife |
| The biodiversity ecosystem umbrella (all repos, one hub) | microsoft/Biodiversity |
| The AI-enabled edge device that runs MegaDetector in the field | microsoft/SPARROW |
Run MegaDetectorV6 quickly¶
The current generation is MegaDetectorV6 (smaller, faster architectures including YOLOv9, YOLOv10, and RT-DETR; the compact variant is roughly 2% of the V5 parameter count at comparable accuracy). Try it without writing code via the Hugging Face Space, or load it through PyTorch-Wildlife:
from PytorchWildlife.models import detection as pw_detection
detection_model = pw_detection.MegaDetectorV6()
Full usage, model selection, and the fine-tuning pipeline are documented in microsoft/MegaDetector.
MegaDetectorV5 and earlier¶
For MegaDetectorV5 weights and the original repository — primarily developed by Dan Morris during his time at Microsoft — see the archive branch of this repository (formerly microsoft/CameraTraps). Dan continues to actively maintain a community fork at agentmorris/MegaDetector, which remains a valuable resource for the community.
Citing MegaDetector¶
A citation.cff is maintained on microsoft/MegaDetector for automated citation tools. Cite Beery, Morris, Yang 2019 — Efficient Pipeline for Camera Trap Image Review for MegaDetector specifically, and Hernandez et al. 2024 — Pytorch-Wildlife for the framework.
Contact¶
Questions about MegaDetector or PyTorch-Wildlife: open an issue on microsoft/MegaDetector or join the Discord.