Fig. 4
From: A comparison of point-tracking algorithms in ultrasound videos from the upper limb
Tracking accuracy for DLC. a Tracking accuracy when using one DLC model per subject (Group 1) versus one model per group of subjects (Group 2). b Tracking errors when increasing the number of training iterations for DLC model across subjects (Group 2). The datapoints for Group 1 come from the average errors in 20 subjects, 6 videos each. The datapoints for Group 2 come from the average errors over 20 randomized video groups among 6 subjects. Error bars represent SEM (*** \(p<0.001\))