DeepLabCut (https://github.com/DeepLabCut/) (Mathis et al., 2018; Nath et al., 2019) was used for tracking body parts of mice in an open field arena or in the rotarod. DeepLabCut 2.1.8.2 (local version on Windows with CPU, using the GUI) and 2.1.10.2 (google colab to train the network) were used using default parameters and the pretrained resnet50 network with imgaug augmentation. Frames were extracted with the k-means method and outlier frames with the jump method. Open field: 20 images from 19 videos (10 or 30 fps) were extracted for a total of 380 labeled pictures. 8 body parts (snout, both ears, body center, both side laterals, tail base and tail end) and the 4 corners of the field arena were manually labeled and linked to each other using skeletons. A neural network was trained using these images for 170K iterations. 20 outlier frames were extracted from each video and relabeled. An additional 20 images from 19 videos with different recording conditions were labeled. The network was then refined for 210K iterations (from scratch), yielding a train error of 3.33 pixels and a test error of 8.83 pixels (with a likelihood p-cutoff of 0.6). This process was repeated a second time (using an additional 20 images from 15 new videos) to improve the pixel error; to a final 400 K iterations (train error: 2.65, test error: 3.71). 67 videos from 5 different experiments were analyzed on the final network. Used to analyze videos for a publication (Labouesse et al., Nature Communications 2023)