Sketch-based image retrieval via Siamese convolutional neural network

Sketch-based image retrieval (SBIR) is a challenging task due to the ambiguity inherent in sketches when compared with photos. In this paper, we propose a novel convolutional neural network based on Siamese network for SBIR. The main idea is to pull output feature vectors closer for input sketch-ima...

Full description

Saved in:
Bibliographic Details
Published inProceedings - International Conference on Image Processing pp. 2460 - 2464
Main Authors Yonggang Qi, Yi-Zhe Song, Honggang Zhang, Jun Liu
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.09.2016
Subjects
Feature extraction
Image edge detection
Image retrieval
Neural networks
SBIR
Shape
Siamese CNN
Training
Visualization
Online AccessGet full text

Cover

More Information
Summary:Sketch-based image retrieval (SBIR) is a challenging task due to the ambiguity inherent in sketches when compared with photos. In this paper, we propose a novel convolutional neural network based on Siamese network for SBIR. The main idea is to pull output feature vectors closer for input sketch-image pairs that are labeled as similar, and push them away if irrelevant. This is achieved by jointly tuning two convolutional neural networks which linked by one loss function. Experimental results on Flickr15K demonstrate that the proposed method offers a better performance when compared with several state-of-the-art approaches.
ISSN:2381-8549
DOI:10.1109/ICIP.2016.7532801