Auto-Supervision Adaptative au Domaine : Une Pipeline Éco-Efficiente pour la Détection d'Anomalies Thoraciques

Read full text Summarise with VIRAT
Authors: Clément Frerebeau, Marc Habib, Sélim Jomaa, Eve Jouni, Bryan Fozame, Doreid Ammar, Anuradha Kar
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Automated interpretation of chest X-rays (CXR) (1) is essential for mass screening, but remains hampered by class imbalance and the domain gap between natural and medical images. While the current state of the art (2025), dominated by massive models such as LungMaxViT (2), maximizes accuracy, these solutions impose prohibitive computational costs, incompatible with resource-constrained clinical deployment (Edge AI).

Fields of Study & Subjects

COVID-19 diagnosis using AI Anomaly Detection Techniques and Applications Domain Adaptation and Few-Shot Learning

Publication details

Journal / source
HAL
Publication date
9 March 2026
Language
FR