Toward few pixel annotations for 3D segmentation of material from electron tomography - Université Jean-Monnet-Saint-Étienne Access content directly
Conference Papers Year :

Toward few pixel annotations for 3D segmentation of material from electron tomography

Cyril Li
  • Function : Author
  • PersonId : 1231731
Christophe Ducottet
Maxime Moreaud

Abstract

Segmentation is a notorious tedious task, especially for 3D volume of material obtained via electron tomography. In this paper, we propose a new method for the segmentation of such data with only few partially labeled slices extracted from the volume. This method handles very restricted training data, and particularly less than a slice of the volume. Moreover, unlabeled data also contributes to the segmentation. To achieve this, a combination of self-supervised and contrastive learning methods are used on top of any 2D segmentation backbone. This method has been evaluated on three real electron tomography volumes.
Fichier principal
Vignette du fichier
VISAPP2023_LI_Toward_few_pixel_annotations_CameraReady.pdf (1.47 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

ujm-04006630 , version 1 (27-02-2023)

Identifiers

  • HAL Id : ujm-04006630 , version 1

Cite

Cyril Li, Christophe Ducottet, Sylvain Desroziers, Maxime Moreaud. Toward few pixel annotations for 3D segmentation of material from electron tomography. International Conference on Computer Vision Theory and Applications, VISAPP 2023, Feb 2023, Lisbonne, Portugal. ⟨ujm-04006630⟩
42 View
32 Download

Share

Gmail Facebook Twitter LinkedIn More