Open-field and NORT scenarios:

• The rat moves freely in a rectangular box, 

• The number of visits and visit duration in specific box area must be computed,

• The YOLOv9 model is used for the segmentation of the scene elements (cage, other objects) and rat's image.

Y-maze scenario:

• The rat moves in a Y-shaped maze,

• The number of visits and visit durations for each branch of the maze must be computed,

• The YOLOv9 model is used for the detection of the rat in each context.

Implementation: the YOLOv9 training, evaluation and detection are implemented in Python, parameters computing and final reports generation in C++. The Python-C API interface is used to run the Python scripts from the main C++ application.