Automated Tracking of Whiskers in Videos of Head Fixed Rodents

摘要: We have developed software for fully automated tracking of vibrissae (whiskers) in high-speed videos (>500 Hz) head-fixed, behaving rodents trimmed to a single row whiskers. Performance was assessed against manually curated dataset consisting 1.32 million video frames comprising 4.5 whisker traces. The current implementation detects whiskers with recall 99.998% and identifies individual 99.997% accuracy. average processing rate these images 8 Mpx/s/cpu (2.6 GHz Intel Core2, 2 GB RAM). This translates 35 processed per second 640 px×352 px 4 speed accuracy achieved enables quantitative behavioral studies where the analysis millions is required. used analyze evolving whisking strategies as mice learned whisker-based detection task over course 6 days (8148 trials, 25 frames) measure forces at sensory follicle that most underlie haptic perception.