Pedestrian or person tracking using 2D LiDAR sensors has significantly piqued the researchers’ interest. While conventional tracking methods successfully track people in a wide room using 2D-LiDAR, they often face occlusion issues due to population density. This situation presents detection challenges, necessitating the suspension of tracking. To get a more accurate tracking of the pedestrians in a frame, we apply a type of transfer learning algorithm, which is a multi-object tracking algorithm named ByteTrack, to the 2D-LiDAR-based ankle tracking system. We have enabled it to identify every step of an individual pedestrian when an occlusion takes place, and we have also enhanced the tracking’s processing speed in every complex scene.