EDNet: Zero-shot classification for ceramic package substrates surface defect with embedding diffusion network

doi:10.1016/j.patcog.2026.113811

EDNet: Zero-shot classification for ceramic package substrates surface defect with embedding diffusion network

2026-04-23

0

PRE

AI

DOI：10.1016/j.patcog.2026.113811

Original

Original request for help

Save

Abstract

En 中文

• This is the first introduction of the zero-shot defect classification for ceramic package substrate. • We have created a novel dataset for zero-shot classification focusing on ceramic package substrate surface. • We proposed an embedding diffusion network which reconstructs the embedding feature and enhances the representation in latent space. • We introduced information mixer and cross-modality integration modules to strengthen the interaction between visual features and semantic features.

Keywords:

zero-shot classification

ceramic package substrate

defect detection

embedding diffusion network

cross-modality integration

EDNet: Zero-shot classification for ceramic package substrates surface defect with embedding diffusion network

Abstract

Journal

Organization