SQUIDer

SQUIDer is an open-source software platform designed for automated detection, tracking, and behavioral analysis of squid in video recordings. It provides a workflow to extract quantitative measures such as swimming speed, trajectory, body orientation, and fin movement from underwater footage, enabling researchers to study locomotion, schooling, and responses to stimuli. The project emphasizes accessibility for marine biology labs and researchers who work with modest computing resources.

Core components include a neural network-based detector trained to identify squid in varied lighting and backgrounds, a multi-object tracker that combines Kalman filtering with data association, and analysis modules that compute metrics and annotate behavioral events. The system supports batch processing of large video sets, multi-camera synchronization when used in field experiments, and export of results to standard formats such as CSV and JSON. It also offers a plugin interface for custom analyses and visualization tools for trajectory plots and heatmaps.

Development and release: SQUIDer originated at the Oceanic Research Institute and entered public release in 2020, with ongoing maintenance and community contributions. The project maintains documentation, tutorials, and example datasets to facilitate adoption by researchers and students. It is distributed under an open-source license.

Applications and impact: Researchers use SQUIDer to quantify locomotor parameters, analyze schooling dynamics, and study behavioral responses to environmental changes. The software supports both laboratory experiments with controlled lighting and field recordings from submersible platforms, and serves as a reproducible workflow for behavioral ecology studies.

Reception and future directions: The tool has been cited in user case studies and is being extended to support related cephalopods and additional behavioral classifiers. Planned enhancements include improved pose estimation, real-time processing capabilities, and tighter integration with image-caption data for metadata management.