*4.2. Qualitative Results*

Figures 16 and 17 are the quantitative results on SSDD and HRSID where we only show the quantitative comparison results with the second-best mask AP.

**Figure 16.** Qualitative results on SSDD. (**a**) PANet with the second-best mask AP. (**b**) GCBANet. Green boxes denote the ground truths. Orange boxes denote the false alarms (i.e., false positives, FP). Red circles denote the missed detections (i.e., false negatives, FN). #i denote the ith picture of results.

**Figure 17.** Qualitative results on HRSID. (**a**) GROIE with the second-best mask AP. (**b**) GCBANet. Green boxes denote the ground truths. Orange boxes denote the false alarms (i.e., false positives, FP). Red circles denote the missed detections (i.e., false negatives, FN). #i denote the ith picture of results.

From Figures 16 and 17, the following conclusions can be drawn:


Given the above, GCBANet achieves state-of-the-art SAR ship instance segmentation performance.
