SEP-YOLO: Fourier-Domain Feature Representation for Transparent Object Instance Segmentation

TL;DR

SEP-YOLO introduces a novel Fourier-domain feature enhancement framework for transparent object segmentation, addressing boundary and contrast challenges with dual-domain collaboration and multi-scale refinement, achieving state-of-the-art results.

Contribution

The paper presents SEP-YOLO, a new framework combining frequency domain enhancement and multi-scale spatial refinement for improved transparent object segmentation.

Findings

Achieves state-of-the-art performance on Trans10K and GVD datasets.

Provides high-quality annotations for the Trans10K dataset.

Effectively enhances boundary details using frequency domain techniques.

Abstract

Transparent object instance segmentation presents significant challenges in computer vision, due to the inherent properties of transparent objects, including boundary blur, low contrast, and high dependence on background context. Existing methods often fail as they depend on strong appearance cues and clear boundaries. To address these limitations, we propose SEP-YOLO, a novel framework that integrates a dual-domain collaborative mechanism for transparent object instance segmentation. Our method incorporates a Frequency Domain Detail Enhancement Module, which separates and enhances weak highfrequency boundary components via learnable complex weights. We further design a multi-scale spatial refinement stream, which consists of a Content-Aware Alignment Neck and a Multi-scale Gated Refinement Block, to ensure precise feature alignment and boundary localization in deep semantic features. We also provide high-quality instance-level annotations for the Trans10K dataset, filling the critical data gap in transparent object instance segmentation. Extensive experiments on the Trans10K and GVD datasets show that SEP-YOLO achieves state-of-the-art (SOTA) performance.