Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL

Abstract: 

Current object segmentation algorithms are based on the hypothesis that one has access to a very large amount of data. In this paper, we aim to segment objects using only tiny datasets. To this extent, we propose a new automatic part-based object segmentation algorithm for non-deformable and semi-deformable objects in natural backgrounds. We have developed a novel shape descriptor which models the local boundaries of an object's part. This shape descriptor is used in a bag-of-words approach for object detection. Once the detection process is performed, we use the background and foreground likelihood given by our trained shape model, and the information from the image content, to define a dense CRF model. We use a mean field approximation to solve it and thus segment the object of interest. Performance evaluated on different datasets shows that our approach can sometimes achieve results near state-of-the-art techniques based on big data while requiring only a tiny training set.

up
0 users have voted:

Paper Details

Authors:
Maxime Tremblay, André Zaccarin
Submitted On:
12 September 2017 - 11:38am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Maxime Tremblay
Paper Code:
MA-PG.1
Document Year:
2017
Cite

Document Files

Poster at ICIP2017

(19 downloads)

Subscribe

[1] Maxime Tremblay, André Zaccarin, "LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1945. Accessed: Nov. 23, 2017.
@article{1945-17,
url = {http://sigport.org/1945},
author = {Maxime Tremblay; André Zaccarin },
publisher = {IEEE SigPort},
title = {LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL},
year = {2017} }
TY - EJOUR
T1 - LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL
AU - Maxime Tremblay; André Zaccarin
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1945
ER -
Maxime Tremblay, André Zaccarin. (2017). LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL. IEEE SigPort. http://sigport.org/1945
Maxime Tremblay, André Zaccarin, 2017. LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL. Available at: http://sigport.org/1945.
Maxime Tremblay, André Zaccarin. (2017). "LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL." Web.
1. Maxime Tremblay, André Zaccarin. LEARNING TO SEGMENT ON TINY DATASETS: A NEW SHAPE MODEL [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1945