Overview
Topic status: We're looking for students to study this topic.
Project Summary
Logos are a key component of how organisational entities are presented in visual media. Logo detection and recognition is therefore important in indexing and data mining in video content. In this type of media the logos may appear in various forms, including as part of a video overlay or as part of the physical scene. Successful logo recognition depends on robust object recognition techniques to be able to deal with the variation of logo appearance across different videos.
This project will implement a logo recognition system for news video. The main tasks include reviewing logo recognition literature, assembling a database of logos and the identities of the entities they are associated with; applying feature extraction techniques to produce a compact and robust representation of each logo; applying clustering techniques to produce a compact visual feature vocabulary and encoding the logo data set using this vocabulary; implementing matching and geometric verification techniques; and conducting experiments on news video to evaluate the performance of the logo recognition system.
Expected outcomes, applications and/or benefits
- An understanding of the process of recognition research, logo recognition systems and object recognition in general.
- A data set and framework for training and testing a logo recognition system.
- A system for detecting and recognising logos in video, both as overlay images or as part of a natural scene.
Required student skills/experience
Strong C++ programming experience.
- Study level
- Vacation research experience scholarship
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Keywords
- video, recognition, logo
- Contact
- Contact the supervisor for more information
Professor Sridha Sridharan
Associate Professor Clinton Fookes
