Fry Tracking and Temp Behavior

Fry Tracking and Behavior Analysis.
The segmentation dataset comprised both side-view and top-view images of 20 fry, captured simultaneously by two cameras over 25 and 52 minutes respectively, ensuring comprehensive visual information for model training. To validate real-time tracking accuracy, we developed a separate dataset featuring varying fry densities (20, 30, 40, 50, and 60 fry) in complex scenarios including crossing paths and occlusions.
Our experiment was structured into three 8-hour phases: cooling from 28°C to 20°C, maintaining a constant 20°C, and reheating from 20°C to 28°C. Temperature changes were implemented gradually, altering by 1°C every ten minutes, using a water circulation system with cooling and heating elements to ensure indirect and stable temperature control.