Overview
MegaDetector is an AI model that identifies animals, people, and vehicles in camera trap images (which also makes it useful for eliminating blank images). This model is trained on several million images from a variety of ecosystems.
Getting started
If you are a biologist looking to get started with MegaDetector, you probably don't want to download the model directly; instead, check out AddaxAI (formerly EcoAssist), a GUI-based tool for running AI models on camera trap images.
If you want to run MegaDetector directly, check out the MegaDetector User Guide.
Variations
Two variations of the model are included: MDv5a and [MDv5b](https://huggingface.co/agentmorris/megadetector/resolve/main/md_v5b.0.0.pt.
MegaDetector v5b was trained only on camera trap images (several million images).
MegaDetector v5a is trained on the same camera trap data as MegaDetector v5b, but it is also trained on images from the COCO dataset.
Both variations use the same architecture (YOLOv5x6).
License information
Both variations were trained with a GPL-licensed version of the YOLOv5 framework. No restrictions are placed on the use of the model by the MegaDetector developers; the applicable license may be determined by the code you use to run the model.
Sample output
Here's a “teaser” image of what MegaDetector output looks like:
Image credit University of Washington.