DC Field | Value | Language |
---|---|---|
dc.contributor.author | Choi, Sung Pill | - |
dc.date.accessioned | 2023-06-23T19:34:01Z | - |
dc.date.available | 2023-06-23T19:34:01Z | - |
dc.date.issued | 2019 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1021101&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/309156 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 전기및전자공학부, 2019.8,[v, 97 p. :] | - |
dc.description.abstract | A low-power 3D hand gesture tracking (HGT) processor is proposed for real-time user interaction in smart mobile devices. It features: 1) 1.80TOPS/W CNN-Stereo core for energy-efficient depth sensing | - |
dc.description.abstract | 2) Triple ping-pong buffers with workload balancing to reduce 23.9% processing time | - |
dc.description.abstract | and 3) Nearest neighbor searching processing-in-memory for hand tracking to achieve 2.8x energy efficiency. The proposed SoC facilitates real-time 3D HGT with 33.5mW and maximally 10.8mm error. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | CNN processor▼a3D hand tracking▼aProcessing in memory▼aMemory efficient hardware▼aLow power | - |
dc.subject | 컨볼루션 신경망 프로세서▼a3D 핸드 트래킹▼a컴퓨팅 메모리▼a메모리 효율적 하드웨어▼a저전력 | - |
dc.title | Low-power real-time CNN-based hand tracking hardware for mobile devices | - |
dc.title.alternative | 모바일 디바이스를 위한 CNN기반 저전력 실시간 핸드 트랙킹 하드웨어 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 최성필 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.