Figure 2
From: A comparative analysis of human and AI performance in forensic estimation of physical attributes
The distribution of in-the-wild height/weight errors for AI, expert, non-expert crowd, non-expert individuals, and baseline. The open circles and horizontal error bars correspond to the median error and \(95\%\) confidence intervals. The two tables denote the pair-wise statistical significance at \(p<0.005\) (\(\star\)) or \(p<0.0005\) (\(\star \star\)) between different groups. See also Table 1.
