A paper developed within the ENCIRCLE project has been accepted for presentation at the 2026 IEEE International Conference on Image Processing (ICIP 2026), to be held in Tampere, Finland, from 13 to 17 September 2026. IEEE ICIP is the flagship IEEE conference on image processing, vision and imaging, organised annually since 1994 and attracting leading researchers and practitioners from around the world.
The accepted paper, titled Improving zero-shot industrial defect detection exploiting LMMs as inverse reasoners, is authored by Maria Tzelepi, Nikolaos Dimitriou, Christos Gogos and Dimitrios Tzovaras from CERTH, the ENCIRCLE project coordinator.
The work addresses the challenge of industrial defect detection in manufacturing environments where labeled defective data is scarce. The authors propose a novel method that uses Large Multimodal Models (LMMs) as inverse reasoners to generate counterfactual defect explanations for normal images. These explanations are then used to improve the defect discrimination ability of a CLIP-based Zero-Shot Anomaly Detection (ZSAD) model, resulting in a significant performance improvement over the state of the art. The method is also extendable to generic detection and classification tasks.
This publication marks an important milestone in ENCIRCLE’s research trajectory, demonstrating that the project’s work on AI-driven quality assurance in industrial manufacturing is producing results of international scientific impact.
More about IEEE ICIP 2026 à https://2026.ieeeicip.org/