BioDetect - Deep Learning for Biodiversity Detection and Classification

Monitoring biodiversity is a fundamental challenge in protecting the natural world. Collecting data on biodiversity and ecosystems with traditional human-based surveys is time-consuming, therefore ecology has increasingly turned to remote image capture. However, the challenge of turning enormous pools of imagery into biologically meaningful information has severely restricted the growth of automated monitoring. This collaborative project will broaden the use of cutting-edge software tools in ecology and evolution. Building from existing datasets in tropical ecosystems and Swiss aquatic environments, we are developing and distributing machine learning and computer vision tools for biodiversity detection and identification. The algorithms developed in this project will be open source and freely shared with the biology and data science communities as well as with the public, to foster growth in the respective disciplines as well as be available for citizen science projects.