Skip to main content
. 2023 May 30;17:1111908. doi: 10.3389/fnbeh.2023.1111908

Figure 9.

Figure 9

AlphaTracker surpasses the performance of both DeepLabCut and SLEAP in tracking two mice in a home cage. (A–C) Show the Average Precision (AP), Multi-Object Tracking Precision (MOTP), and Multi-Object Tracking Accuracy (MOTA) metrics for different keypoints of AlphaTracker results. The metrics were evaluated for different amounts of training frames (25, 50, 100, 200, 300, and 600 frames) using a 200-frame evaluation dataset. Different colors represent different keypoints including snouts (blue), ears (yellow) and tailbases (green) and the total metrics (red). (D–F) Show the evaluation results for AlphaTracker (red), SLEAP (green), and DeepLabCut (blue), with connected dots representing the total metric and the unconnected dots representing the metric for individual body parts.