Guy Shani

Continuous trajectories of small insects can provide measurements of movement distance, speed, resting periods, spatial preference, and contact with plant material, but manual observation is labor-intensive. We present a low-cost smartphone-to-web workflow for aphid tracking in controlled laboratory experiments. The system integrates image acquisition, online upload and experiment management, training-free detection using multi-threshold binarization and blob filtering, Kalman-filter prediction, global nearest-neighbor association, trajectory visualization, and data export. The contribution is the accessible end-to-end integration of established methods rather than a new detection or tracking algorithm. Across nine controlled sequences, the mean precision, recall, and Multiple Object Tracking Accuracy (MOTA) were 0.852, 0.845, and 0.671, respectively, with MOTA ranging from 0.122 to 0.915. Performance was highest under clean, well-focused conditions and degraded in the presence of blur, dirt, and aphid-like stationary objects. The current implementation assumes a standardized overhead view and a light, visually uniform background. field deployment and tracking accuracy over multi-hour or multi-day sequences were not evaluated.