SAM-PTx: Text-Guided Fine-Tuning of SAM with Parameter-Efficient, Parallel-Text Adapters

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jalilian, Shayan, Bais, Abdul
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913969428496384
author Jalilian, Shayan
Bais, Abdul
author_facet Jalilian, Shayan
Bais, Abdul
contents The Segment Anything Model (SAM) has demonstrated impressive generalization in prompt-based segmentation. Yet, the potential of semantic text prompts remains underexplored compared to traditional spatial prompts like points and boxes. This paper introduces SAM-PTx, a parameter-efficient approach for adapting SAM using frozen CLIP-derived text embeddings as class-level semantic guidance. Specifically, we propose a lightweight adapter design called Parallel-Text that injects text embeddings into SAM's image encoder, enabling semantics-guided segmentation while keeping most of the original architecture frozen. Our adapter modifies only the MLP-parallel branch of each transformer block, preserving the attention pathway for spatial reasoning. Through supervised experiments and ablations on the COD10K dataset as well as low-data subsets of COCO and ADE20K, we show that incorporating fixed text embeddings as input improves segmentation performance over purely spatial prompt baselines. To our knowledge, this is the first work to use text prompts for segmentation on the COD10K dataset. These results suggest that integrating semantic conditioning into SAM's architecture offers a practical and scalable path for efficient adaptation with minimal computational complexity.
format Preprint
id arxiv_https___arxiv_org_abs_2508_00213
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SAM-PTx: Text-Guided Fine-Tuning of SAM with Parameter-Efficient, Parallel-Text Adapters
Jalilian, Shayan
Bais, Abdul
Computer Vision and Pattern Recognition
Machine Learning
The Segment Anything Model (SAM) has demonstrated impressive generalization in prompt-based segmentation. Yet, the potential of semantic text prompts remains underexplored compared to traditional spatial prompts like points and boxes. This paper introduces SAM-PTx, a parameter-efficient approach for adapting SAM using frozen CLIP-derived text embeddings as class-level semantic guidance. Specifically, we propose a lightweight adapter design called Parallel-Text that injects text embeddings into SAM's image encoder, enabling semantics-guided segmentation while keeping most of the original architecture frozen. Our adapter modifies only the MLP-parallel branch of each transformer block, preserving the attention pathway for spatial reasoning. Through supervised experiments and ablations on the COD10K dataset as well as low-data subsets of COCO and ADE20K, we show that incorporating fixed text embeddings as input improves segmentation performance over purely spatial prompt baselines. To our knowledge, this is the first work to use text prompts for segmentation on the COD10K dataset. These results suggest that integrating semantic conditioning into SAM's architecture offers a practical and scalable path for efficient adaptation with minimal computational complexity.
title SAM-PTx: Text-Guided Fine-Tuning of SAM with Parameter-Efficient, Parallel-Text Adapters
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2508.00213