Fig. 2

The processing pipeline of the dataset, estimated body parts by DeepLabCut, and FOV registration across animals and sessions. (a) The analytical pipeline. FOV, field of view; ROI, region of interest. (b–d) Keypoints in videos of the upper body (b), right side of the face (c), and right eye (d), detected by DeepLabCut. These keypoints are also summarized in Table 2. (e) Schematic illustration of atlas registration to Allen CCF using MesoNet and affine transformations (see Methods).