Mobile visual product search

• Main Author: Yang, Zhongxiu
• Other Authors: Yap Kim Hui
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/77741
• Institution: Nanyang Technological University

With the rapid development of mobile visual search technology, image recognition becomes a hot topic for development and research. Nowadays, mobile phone is an indispensable gadget in people’s life. This encourages mobile app developer to create apps for users to obtain information from internet immediately and easily. Currently, there has a large amount of mature software application for mobile image recognition. The purpose of the project is to evaluate the performance of popular image recognition techniques through comparison of recognition accuracy by the label image of sauce bottle under different conditions of illumination, occlusion, resolution and angles. In the project, the database consists 12 categories label image of sauce bottle which is taking by mobile phone, and some suitable techniques will be proceeding such as bog-of-word (BoW) representation, Scale Invariant Feature Transform (SIFT) descriptor, histogram representation and sparse representation (SRC). From the results, bag-of-word with SIFT method perform the better recognition accuracy result as compared with sparse representation method of this project. Some unwanted features which produce noises in image will affect the accuracy result in this project. And the performance will be improved when extend the reference image database of this project. Finally, discuss the topics of optimizing image database deeply and improving image recognition efficiency for future analysis.