Make It Up: Fake Images, Real Gains in Generalized Few-shot Semantic Segmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xie, Guohuan, He, Xin, Fan, Dingying, Zhang, Le, Cheng, Ming-Ming, Liu, Yun
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917365655011328
author Xie, Guohuan
He, Xin
Fan, Dingying
Zhang, Le
Cheng, Ming-Ming
Liu, Yun
author_facet Xie, Guohuan
He, Xin
Fan, Dingying
Zhang, Le
Cheng, Ming-Ming
Liu, Yun
contents Generalized few-shot semantic segmentation (GFSS) is fundamentally limited by the coverage of novel-class appearances under scarce annotations. While diffusion models can synthesize novel-class images at scale, practical gains are often hindered by insufficient coverage and noisy supervision when masks are unavailable or unreliable. We propose Syn4Seg, a generation-enhanced GFSS framework designed to expand novel-class coverage while improving pseudo-label quality. Syn4Seg first maximizes prompt-space coverage by constructing an embedding-deduplicated prompt bank for each novel class, yielding diverse yet class-consistent synthetic images. It then performs support-guided pseudo-label estimation via a two-stage refinement that i) filters low-consistency regions to obtain high-precision seeds and ii) relabels uncertain pixels with image-adaptive prototypes that combine global (support) and local (image) statistics. Finally, we refine only boundary-band and unlabeled pixels using a constrained SAM-based update to improve contour fidelity without overwriting high-confidence interiors. Extensive experiments on PASCAL-$5^i$ and COCO-$20^i$ demonstrate consistent improvements in both 1-shot and 5-shot settings, highlighting synthetic data as a scalable path for GFSS with reliable masks and precise boundaries.
format Preprint
id arxiv_https___arxiv_org_abs_2603_27206
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Make It Up: Fake Images, Real Gains in Generalized Few-shot Semantic Segmentation
Xie, Guohuan
He, Xin
Fan, Dingying
Zhang, Le
Cheng, Ming-Ming
Liu, Yun
Computer Vision and Pattern Recognition
Generalized few-shot semantic segmentation (GFSS) is fundamentally limited by the coverage of novel-class appearances under scarce annotations. While diffusion models can synthesize novel-class images at scale, practical gains are often hindered by insufficient coverage and noisy supervision when masks are unavailable or unreliable. We propose Syn4Seg, a generation-enhanced GFSS framework designed to expand novel-class coverage while improving pseudo-label quality. Syn4Seg first maximizes prompt-space coverage by constructing an embedding-deduplicated prompt bank for each novel class, yielding diverse yet class-consistent synthetic images. It then performs support-guided pseudo-label estimation via a two-stage refinement that i) filters low-consistency regions to obtain high-precision seeds and ii) relabels uncertain pixels with image-adaptive prototypes that combine global (support) and local (image) statistics. Finally, we refine only boundary-band and unlabeled pixels using a constrained SAM-based update to improve contour fidelity without overwriting high-confidence interiors. Extensive experiments on PASCAL-$5^i$ and COCO-$20^i$ demonstrate consistent improvements in both 1-shot and 5-shot settings, highlighting synthetic data as a scalable path for GFSS with reliable masks and precise boundaries.
title Make It Up: Fake Images, Real Gains in Generalized Few-shot Semantic Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.27206