In this experiment, adult male mice C57BL/6J (aged 2-3 months at the start of the experiment) were given free access to food and water and lived in a 12-hour light/diurnal cycle (lights turned on at 7 am). In the experiment, we conducted experiments in a circular swimming pool measuring 120 cm in diameter and 40 cm in height. The pool contained opaque water (maintained at 20 ± 1°C) achieved by adding titanium dioxide. The mice were trained to escape onto a hidden platform positioned 30 cm away from the pool wall and 1 cm below the water surface. Visual cues were placed on the pool wall to provide spatial references for the platform’s location. Importantly, the platform’s position remained fixed throughout all the training trials.47
For the creation of our training and validation set, we utilized 6 experimental videos with a combined duration of 3 minutes and 38 seconds. From these videos, we extracted a total of 1392 images to create our dataset (training and validation) for 3 key-points. In the test dataset, we used a 20-second video comprising 500 frames of images to compare and evaluate the final recognition results. The lengths of training and testing datasets were determined based on the course of specific mouse behavior in the water maze experiments.
In the water maze experiment (Figure 3C, bottom), we encountered challenges due to poor lighting conditions and the maze partially obstructing the mouse’s body. Specifically, when the mouse turned, its head became completely immersed in water. These factors posed difficulties for testing the reliability and accuracy of our pose estimation method.
Do you have any questions about this protocol?
Post your question to gather feedback from the community. We will also invite the authors of this article to respond.