Machine Learning for Conservation

Animal Detection Network

Mission: Advance the use of deep machine learning for automated identification and counting of animal species in images, video, and audio.

Initiative Goals

Flagship Project: Species Identification and Localization in Camera Trap Images

Project Goals

  • Develop opensource tools for annotating and packaging data for training and evaluating deep machine learning pipelines aimed at identifying animal species in images from camera traps (i.e, trail cameras).
  • Steward an open, labeled dataset of camera trap imagery for training and evaluating deep machine learning algorithms.
  • Develop and host a model testing portal to allow researchers and developers to run their models against a curated, non-public dataset to allow direct comparison of machine learning workflows for identifying animal species in camera trap images.
  • Engage conservation practitioners and citizen scientists to identify partners who will commit to a species and provide curated sets of annotated data for inclusion and distribution in a broader open, labeled dataset.
  • Provide pretrained deep machine learning models, workflows and guides to introduce conservation practitioners to the utility of deep machine learning for automated identification and counting of animals.
  • Cultivate a network of conservation practitioners to advance the development and use of deep machine learning for automated identification, counting, and monitoring of animals by hosting or convening working groups.

Detailed Project Description

Emerging Project: Localization and counting of animals in low level aerial images

This project will develop an advanced image analysis pipeline to detect, identify, and count animals in digital, low alitide, aerial images.



Contact Us

If you have any questions or are interested in this project please contact Ned Horning (Director of Applied Biodiversity Informatics, Center for Biodiversity and Conservation ) and Peter Ersts (Software Developer, Center for Biodiversity and Conservation) .

X
Email:


First & Last Name:


Question / Comment:


Add me to your contact list: