Fig. 3
From: A fine-tuned foundational model SurgiSAM2 for surgical video anatomy segmentation and detection
Performance comparison of Hiera Large and Hiera Base Plus backbones using weighted mean Dice coefficient (x-axis), evaluated with 1 to 10 prompt points (y-axis) across segmentation tasks. The performance of Hiera-Large and Hiera-Base-Plus increases with progressively increasing number of prompts plateauing near the higher number of prompts.
