Title :
A 86mW 98GOPS ANN-searching processor for Full-HD 30fps video object recognition with zeroless locality-sensitive hashing
Author :
Kim, Gyeonghoon ; Oh, Jinwook ; Yoo, Hoi-Jun
Author_Institution :
Dept. of EE, KAIST, Daejeon, South Korea
Abstract :
Increasing database size and massive dimensions of keypoint descriptors caused nearest neighbor searching as a main bottleneck in object recognition systems. Therefore a high throughput approximate nearest neighbor (ANN) searching processor is proposed for real-time object recognition. To reduce the external bandwidth required in nearest neighbor searching, this chip utilizes an on-chip cache for transaction reduction and the zeroless locality-sensitive hashing (zeroless-LSH) operation for required data suppression. As a result, the proposed ANN-searching processor achieves 62,720 vectors/sec throughput and 1,140GOPS/W power efficiency, which are 1.45x and 1.37x higher than the state-of-the-art respectively, enabling real-time object recognition for Full-HD 30fps video streams.
Keywords :
neural nets; object recognition; video signal processing; video streaming; ANN searching processor; ANN-searching processor; data suppression; database size; full-HD video object recognition; full-HD video streams; high throughput approximate nearest neighbor; keypoint descriptors; massive dimensions; nearest neighbor searching; object recognition systems; on-chip cache; power efficiency; real-time object recognition; transaction reduction; zeroless locality-sensitive hashing; Artificial neural networks; Indexes; Object recognition; Real-time systems; System-on-a-chip; Vectors;
Conference_Titel :
ESSCIRC (ESSCIRC), 2012 Proceedings of the
Conference_Location :
Bordeaux
Print_ISBN :
978-1-4673-2212-6
Electronic_ISBN :
1930-8833
DOI :
10.1109/ESSCIRC.2012.6341352
