Intelligent Multicore Processor for Attention-based Object Recognition

June 21, 2011
Jimwook Oh | KAIST

In order to fulfill today’s object recognition requirements in mobile environment, the brain-inspired Unified Visual Attention Model (UVAM) is applied to SIFT-based recognition for high processing speed with limited energy budget. In addition, a heterogeneous multicore processor is proposed using SIMD/MIMD/ASIC combined architecture for different data-level parallelism of proposed algorithm. Especially, we integrated an artificial intelligence accelerator for inference and learning operation to enable smart object recognition system for future video applications.

Speaker Details

Jinwook Oh (S’08) is currently working toward the Ph.D. degree in electrical and computer science from the Korea Advanced Institute of Science and Technology (KAIST), Dajeon, Korea in 2010. Since 2008, he has been involved with several System-on-Chip designs, including artificial intelligence realization for high performance and low power chip multiprocessor. He is now working on heterogeneous multicore processor for object recognition on high resolution video images.