LEVERAGING ARTIFICIAL INTELLIGENCE, WEB APP BUILDING PLATFORMS, AND HUMAN VALIDATION TO INCREASE EFFICIENCY OF PROCESSING TRAIL CAMERA IMAGES

Kaitlin R McGee; California Department of Fish and Wildlife; kaitlin.mcgee@wildlife.ca.gov; Bergen Foshay, Matt Toenies, Ryan Peek, Allison Salas, Lindsey Rich, Kaitlin McGee Bergen Foshay

California Department of Fish and Wildlife biodiversity monitoring projects like the California Environmental Monitoring and Assessment Framework and Sentinel Sites for Nature, use trail cameras to collect data on small to large-sized mammals and reptiles at over 300 sites annually, resulting in millions of wildlife camera images each year. To process these data in a timely and consistent manner, we developed a data processing pipeline that uses artificial intelligence (AI), human validation, and sharing information through online platforms. We use Wildlife Insights to store and identify photos because the AI tool automatically identifies blank images and photographed animals to species, reducing our processing time by ~75%. To help people correct or confirm AI-generated species tags and ensure consistency, we created identification guides in ArcGIS Experiences. We provide example camera trap images and descriptions for all photographed animals, including small mammals and reptiles that can be challenging to identify. Our pipeline enables a small team of people, regardless of experience, to quickly process millions of images, making our methods scalable and open to collaboration. It also enables us to ask statewide questions that improve our understanding of how land use and climate-related changes are impacting our state’s natural resources.

Poster Session  

Speaker Bio:

Kaitlin McGee (she/her) is an Environmental Scientist at the California Department of Fish and Wildlife, where she implements state-wide terrestrial monitoring as part of the California Environmental Monitoring and Assessment Framework (CEMAF). With a B.S. from University of California, Davis and a M.S. from Colorado State University, she is interested in interspecific interactions, land use, and forest ecology. In her spare time, she enjoys reading, watching movies, and getting outdoors. At the conference she will be presenting the data processing pipeline CEMAF uses to process camera trap images, including leveraging artificial intelligence and app building platforms.

Bergen Foshay (she/her) is an Environmental Scientist at the California Department of Fish and Wildlife who is also part of CEMAF. She has a B.S. in Aquatic Biology from the University of California, Santa Barbara and is interested in landscape ecology and land management. When not outdoors, she enjoys knitting and reading.