Object segmentation in depth maps with one user click and a synthetically trained fully convolutional network

Matthieu Grard 1, 2 Romain Brégier 1, 3 Florian Sella 1 Emmanuel Dellandréa 2 Liming Chen 2
2 imagine - Extraction de Caractéristiques et Identification
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
3 IMAGINE - Intuitive Modeling and Animation for Interactive Graphics & Narrative Environments
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : With more and more household objects built on planned obsolescence and consumed by a fast-growing population, hazardous waste recycling has become a critical challenge. Given the large variability of household waste, current recycling platforms mostly rely on human operators to analyze the scene, typically composed of many object instances piled up in bulk. Helping them by robotizing the unitary extraction is a key challenge to speed up this tedious process. Whereas supervised deep learning has proven very efficient for such object-level scene understanding, e.g., generic object detection and segmentation in everyday scenes, it however requires large sets of per-pixel labeled images, that are hardly available for numerous application contexts, including industrial robotics. We thus propose a step towards a practical interactive application for generating an object-oriented robotic grasp, requiring as inputs only one depth map of the scene and one user click on the next object to extract. More precisely, we address in this paper the middle issue of object seg-mentation in top views of piles of bulk objects given a pixel location, namely seed, provided interactively by a human operator. We propose a twofold framework for generating edge-driven instance segments. First, we repurpose a state-of-the-art fully convolutional object contour detector for seed-based instance segmentation by introducing the notion of edge-mask duality with a novel patch-free and contour-oriented loss function. Second, we train one model using only synthetic scenes, instead of manually labeled training data. Our experimental results show that considering edge-mask duality for training an encoder-decoder network, as we suggest, outperforms a state-of-the-art patch-based network in the present application context.
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01674511
Contributeur : Matthieu Grard <>
Soumis le : mercredi 3 janvier 2018 - 08:41:05
Dernière modification le : mercredi 11 avril 2018 - 01:57:56
Document(s) archivé(s) le : mercredi 4 avril 2018 - 12:37:17

Fichiers

main.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01674511, version 1
  • ARXIV : 1801.01281

Citation

Matthieu Grard, Romain Brégier, Florian Sella, Emmanuel Dellandréa, Liming Chen. Object segmentation in depth maps with one user click and a synthetically trained fully convolutional network. 2018. 〈hal-01674511〉

Partager

Métriques

Consultations de la notice

340

Téléchargements de fichiers

111