Title: SIFT Matching with CNN Evidences for Particular Object Retrieval
|
| Authors: Zhang, GX; Zeng, Z; Zhang, SW; Zhang, Y; Wu, WC
|
| Author Full Names: Zhang, Guixuan; Zeng, Zhi; Zhang, Shuwu; Zhang, Yuan; Wu, Wanchun
|
| Source: NEUROCOMPUTING, 238 399-409; 10.1016/j.neucom.2017.01.081 MAY 17 2017
|
| Language: English
|
| Abstract: Many object instance retrieval systems are typically based on matching of local features, such as SIFT. However, these local descriptors serve as low-level clues, which are not sufficiently distinctive to prevent false matches. Recently, deep convolutional neural networks (CNN) have shown their promise as a semantic-aware representation for many computer vision tasks. In this paper, we propose a novel approach to employ CNN evidences to improve the SIFT matching accuracy, which plays a critical role in improving the object retrieval performance. To weaken the interference of noise, we extract compact CNN representations from a number of generic object regions. Then a query-adaptive method is proposed to choose appropriate CNN evidence to verify each pre-matched SIFT pair. Two different visual matching verification functions are introduced and evaluated. Moreover, we investigate the suitability of fine-tuning the CNN for our proposed approach. Extensive experiments on benchmark dataSets demonstrate the effectiveness of our method for particular object retrieval. Our results compare favorably to the state-of-the-art methods with acceptable memory usage and query time. (C) 2017 Elsevier B.V. All rights reserved.
|
| ISSN: 0925-2312
|
| eISSN: 1872-8286
|
| IDS Number: EP4TF
|
| Unique ID: WOS:000397372100035
*Click Here to View Full Record
