New SAMSEM approach adapts SAM2 for IC metal line segmentation

Researchers have developed SAMSEM, a novel approach for segmenting metal lines in integrated circuit (IC) images, adapting Meta's Segment Anything Model 2 (SAM2). This method utilizes a multi-scale segmentation strategy and incorporates topology-based losses alongside pixel-based ones to focus on electrical connectivity. SAMSEM demonstrates strong generalization capabilities across various ICs, technology nodes, and manufacturing processes, achieving low error rates even on unseen data. AI

IMPACT Enhances hardware security and verification by improving the accuracy and generalizability of IC metal line segmentation.

RANK_REASON This is a research paper detailing a new method for IC metal line segmentation using an adapted foundation model. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• integrated circuit
• Julian Speith
• Meta
• SAM2
• Segment Anything Model 2

• paper
• product
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →