Extended Data Fig. 2: Performance comparison of SBeA and SLEAP. | Nature Machine Intelligence

Extended Data Fig. 2: Performance comparison of SBeA and SLEAP.

From: Multi-animal 3D social pose estimation, identification and behaviour embedding with a few-shot learning framework

Extended Data Fig. 2

a, Prediction error comparison of all test data (two-way ANOVA followed by Sidak multiple comparisons test, N = 14400, adjusted P values from Nose to Tip Tail are <0.0001, <0.0001, <0.0001, 0.8054, <0.0001, <0.0001, 0.0030, <0.0001, 0.4651, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, and <0.0001). b, Prediction error comparison of close contact (two-way ANOVA followed by Sidak multiple comparisons test, N = 4602, adjusted P values from Nose to Tip Tail are <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, <0.0001, >0.9999, <0.0001, <0.0001, and <0.0001). Stems represent the mean values of each violin plot. RMSE: root-mean-squared error, n.s.: no significant difference, *: P < 0.05, **: P < 0.01, ***: P < 0.001, ****: P < 0.0001.

Source data contains the comparison between SBeA and SLEAP.

Back to article page