Development of cloud-based visual image search platform

• Main Author: Nguyen, Kyle Huu Thanh
• Other Authors: Hoi Chu Hong
• Format: Final Year Project
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Special-purpose and application-based systems
• Online Access: http://hdl.handle.net/10356/59076
• Institution: Nanyang Technological University

Ecommerce has evolved to be a vibrant industry with hundreds of billions of dollars in revenue each year. The potential has attracted vast number of businesses into the industry and each of them is looking for a competitive advantage that can separate their businesses from the rest. Seizing the opportunity, a system called Visebuy has been under development to add more values to ecommercial businesses. Visebuy system provides a full stack solution that enables businesses to integrate image search features to the ecommercial websites and mobile applications. In particular, customers of those businesses will have a totally different way of searching for products that they are interested in - They can submit a picture of the product they like and the internal system will be able suggest the available products which are visually similar. Visebuy system is segmented into different layers including Ecommerce Layer, API Layer, Image Processing Layer and Data Storage Layer. Each layer will make use of the services provided by the lower layers to support its upper layers. This divide-and-conquer architecture design has made a clear distinction in term of functionality amongst layers which subsequently enables their independent implementation and development. In this report, we demonstrate the design and implementation of the Application Programming Interface of the Visebuy system to support the ecommercial developers to integrate this niched function into their very own system. In order to bring about clear explanation on the matter, other parts of Visebuy system are also briefly mentioned. The offered services of API Layer fall into two different categories of administrative services and public services. The administrative service is to be used exclusively by Visebuy system for management purposes. The public services include 1) index images to Database, query images that look similar to a given one, delete an image to exclude it from future querying result and remote index to support batch-indexing millions of images. Those public services are expected to provide developers with essential tools to implement a local image search engine. To provide the services, API Layer actually works with different modules of the system such as Databse for storing and retrieving internal information, Data Storage for storing images and Image Processing Layer for indexing and querying images. At the heart of API Layer is the ServiceManager. All the services for communication with Image Processing Layer are created, monitored and managed by this entity. As such, it contributes greatly to the process of keeping all the services and connections with Image Processing Layer under control and ensure stability of the system. The system has constantly been under development and improvement where our approach is kept refined and reviewed. There are still challenges and obstacles ahead that will required further hard work and effort. However, the system has been tested to work with essentially functions while leaving room for future scalability and improvements. This is a proven evidence that we are on the right track in development of the project.