Contrastive learning and physics oriented evaluation for advanced segmentation in electron tomography

• A full formalization of both Mixed Supervised / Self-Supervised Contrastive Learning Segmentation (M3S-CLS) method and the semi-supervised setting that model the partial annotation process provided by the expert. • A physics-oriented evaluation of M3S-CLS with a study of the influence of the annotation ratio, i.e. the surface of the slice annotated by the expert, in the semi-supervised setting. • A comparison of M3S-CLS with the standard Cross-Entropy-based Learning (CEL) in the semi-supervised setting. • A dataset of five fully segmented ET volumes with the corresponding baseline results for both M3S-CLS and CEL. Deep learning methods are now achieving strong results for segmentation tasks, and the standard metric for evaluating methods is the Intersection over Union (IOU). However, we show in this paper that IOU is not efficient in evaluating the quality of segmentation for electron tomography (ET) images of zeolites. We perform a physics-oriented evaluation to ensure that the segmentation results yield coherent physical measures. We also formalize Mixed Supervised / Self-Supervised Contrastive Learning Segmentation (M3S-CLS), a semi-supervised approach using a contrastive learning approach that uses expert annotations to train the neural network model. A detailed comparison of this method with a standard cross-entropy-based model is provided. In addition, we publish a database of five fully segmented ET volumes along with corresponding baseline results. The code and the database is available at https://gitlab.univ-st-etienne.fr/labhc-iscv/M3S-CLS .