FoxMask

FoxMask automates detection of animal presence in camera trap image sequences to enable efficient wildlife monitoring and ecological data collection.

Key Features:

• Automatic Detection: Uses background estimation and foreground segmentation algorithms to identify moving objects likely to be animals within image sequences.
• High Accuracy: Demonstrated classification accuracy of over 90% on a challenging arctic fox den dataset for labeling images as containing an animal or not, thereby reducing false triggers.
• Adaptability to Variable Conditions: Validated on variable conditions including different shapes and colors of arctic foxes against diverse backgrounds during snowmelt and the summer growing season.

Scientific Applications:

• Wildlife monitoring: Automated presence/absence detection in camera trap images to support population and behavioral studies.
• Large-scale dataset processing: Enables rapid screening of extensive camera-trap image collections to increase sample sizes for ecological analyses.
• Elusive and nocturnal species studies: Reduces manual review effort for species that are difficult to observe directly, facilitating studies of cryptic or low-observation taxa.

Methodology:

Background estimation to differentiate static and dynamic elements within image sequences and foreground segmentation to isolate potential animal movements, with parameters tuned to handle variability in wildlife imagery.