Automating the Detection of Animals in Wildlife Cameras – Center for Data Innovation

The California Institute of Technology has published iWildCam, a dataset of nearly 300,000 images from 143 locations to foster the development of AI systems that can automatically detect animals in stationary wildlife cameras. The dataset includes 20 different animal classes, including bobcat, moose, and mountain goat. Biologists use wildlife cameras to monitor the biodiversity and population density of animal species, but automating the process can be difficult because images of the animals can be poorly illuminated, suffer from motion blur, and have distorted perspectives if the animals are too close to the camera. This dataset could make it easier to train AI systems to automate this process.