Fig. 7.

Performance comparison of tested models with all metrics evaluated on the testing set of the Cornell MS dataset. Statistical significance test between of our method and the other state-of-the-art methods were evaluated using a paired t-test. The threshold of the significance was $\alpha =0.05$, and the p-values in the figure are annotated as: * for p < 0.05, ** for p < 0.01, *** for p < 0.001, **** for p < 0.0001, and ns for non-significant.