Continuous RGB-D tracking enables assessment of activity, space use, and group synchrony in group-housed grower pigs under commercial conditions

Housing systems for grower–finisher pigs typically allocate functional zones, with solid floors intended for resting and slatted areas for elimination and locomotion, yet detailed evidence on 24-h space use in commercial pens remains limited. We developed and evaluated a depth-assisted RGB-D tracking pipeline for continuous monitoring of group-housed grower pigs under commercial farm conditions. Twelve pigs in one pen were recorded for 72 h using an overhead Intel RealSense L515 camera. YOLOv8-based RGB detection, depth-assisted post-detection refinement, and DeepSORT-based tracking produced individual trajectories used to compute an Activity Index (AI), Occupancy–Coverage Index (OCI), and group synchrony. Compared with RGB-only tracking, depth-assisted RGB-D tracking improved identity stability, reducing ID switches from 42 to 18 and increasing HOTA from 76.8% to 84.2%. Trajectory-derived analyses captured clear activity peaks during scheduled feeding and cleaning, modest but consistent day–night differences in filtered displacement, non-uniform pen use, and variation among individual tracks. Occupancy heatmaps showed higher use of wall-adjacent areas and lower use of the central passage. These results show that continuous depth-assisted tracking can transform commercial video recordings into interpretable indicators of movement, space use, and group coordination. Because the study was limited to one pen, 12 pigs, and 72 h of recording, the findings should be interpreted as proof-of-concept evidence requiring validation across additional pens and farms.