Preprint
Aug 2026
FIRM: Fine-Grained Intra-Token Representation of Masks for Remote Sensing Reasoning Segmentation
FIRM, a Fine-grained Intra-token Representation of Masks, achieves leading results, including $70.5/80.5/80.5$ gIoU/cIoU on LaSeRS and a $3.0-point average gain on EarthReason.
Weidong Tang, Kai-Yu Li, Yi-Kai Wang et al.
· 0 citations