A Holistic Approach to Robust and Scalable Object Recognition

This project is conducted at Advanced Digital Sciences Center (ADSC), as part of the Interactive Digital Media (IDM) sub-program. The PI leading our research efforts is Dr. Yi Ma from UIUC and MSRA.


[Left plot] Training acquisition system: Four projectors and two cameras controlled by one computer. [Top-right plot] 38 traning images of one subject captured by the training acquisition system. [Bottom-right plot] Representative testing examples, including indoor/outdoor images (1st row), images of subjects with eyeglasses (2nd row), and image of subjects with sunglasses (3rd row).


Except for the applications in augmented reality, object recognition has become the core to many new consumer products and commercial services over the Internet or cellular network. For instance, it can be applied to adult content filtering in the web, traffic surveilance, security access control, visual geolocalization, visio-conferences or intelligent man-machine communication. Moreover, object recognition can be marked as representative killer applications for future cloud computing by IT giants like Google, Microsoft, and Apple. As all these companies have started to realize, effective object recognition methods could bring many disruptive technologies that can fundamentally change public safety, human computer interface, social networking and hence bring tremendous new business models and opportunities.

Description: Figure1_2_input  Description: Image3_1_input Description: Figure1_4_input_crop

Representative examples of objects for recognition: human faces, texts, bar codes, and landmarks.

Although we have listed only a few examples above, people can already get a sense of the broad range of exciting applications and business opportunities that object recognition can potentially bring (or has already brought). These applications would significantly enhance our ability in recognizing new things around us, effectively making the entire Internet as our backup memory and knowledge base. This obviously will fundamentally alter how we will interact with the environment, with other people, and with machines in the future.