PulseAugur
EN
LIVE 22:23:41

New method uses SAM and diffusion models for weakly-supervised object detection

By PulseAugur Editorial · [2 sources] ·

Researchers have developed a new method for weakly-supervised RGB-D Salient Object Detection (SOD) that utilizes the Segment Anything Model (SAM) to generate pseudo annotations from sparse scribbles. This approach, named SAM-PAG, expands these scribbles into dense pixel-level annotations. The generated annotations are then used with a diffusion model called $S^2$Diff, which refines noisy saliency maps by integrating cross-modal features and mitigating noise interference. This combined method achieves competitive performance compared to fully-supervised techniques on multiple datasets. AI

IMPACT This research advances weakly-supervised learning techniques for computer vision tasks, potentially reducing the need for extensive manual annotation in object detection.

RANK_REASON The cluster contains a research paper detailing a new method for salient object detection.

Read on arXiv cs.CV →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New method uses SAM and diffusion models for weakly-supervised object detection

COVERAGE [2]

  1. arXiv cs.CV TIER_1 English(EN) · Wenqi Si, Gongyang Li, Shixiang Shi, Weisi Lin ·

    Weakly-Supervised RGB-D Salient Object Detection via SAM-driven Pseudo Annotation and State Space Interaction-based Diffusion

    arXiv:2607.15041v1 Announce Type: new Abstract: Weakly-supervised RGB-D Salient Object Detection (SOD) is explored to reduce the heavy burden of pixel-level annotations. But scribble annotations lack the structure and details of objects, resulting in inaccurate saliency maps. In …

  2. arXiv cs.CV TIER_1 English(EN) · Weisi Lin ·

    Weakly-Supervised RGB-D Salient Object Detection via SAM-driven Pseudo Annotation and State Space Interaction-based Diffusion

    Weakly-supervised RGB-D Salient Object Detection (SOD) is explored to reduce the heavy burden of pixel-level annotations. But scribble annotations lack the structure and details of objects, resulting in inaccurate saliency maps. In this paper, we propose a novel scribble-supervis…

RELATED ENTITIES

RELATED TOPICS