TY - GEN
T1 - ESRCNN
AU - Jung, Youngbeom
AU - Choi, Yeongjae
AU - Sim, Jaehyeong
AU - Kim, Lee Sup
N1 - Funding Information:
This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NO. 2017R1A2B2009380).
Publisher Copyright:
© 2019 IEEE.
PY - 2019/11
Y1 - 2019/11
N2 - CNN-based Super-Resolution (SR), the most representative of low-level vision task, is a promising solution to improve users' QoS on IoT devices that suffer from limited network bandwidth and storage capacity by effectively enhancing image/video resolution. Although prior accelerators to embed CNN show tremendous performance and energy efficiency, they are not suitable for SR tasks regarding off-chip memory accesses. In this work, we present eSRCNN, a framework that enables performing energy-efficient SR tasks on diverse embedded CNN accelerators by decreasing off-chip memory accesses. To reduce off-chip memory accesses, our framework consists of three steps: a network reformation using a cross-layer weight scaling, a precision minimization with priority-based quantization, and an activation map compression exploiting a data locality. As a result, the energy consumption of off-chip memory accesses is reduced up to 71.89% with less than 3.52% area overhead.
AB - CNN-based Super-Resolution (SR), the most representative of low-level vision task, is a promising solution to improve users' QoS on IoT devices that suffer from limited network bandwidth and storage capacity by effectively enhancing image/video resolution. Although prior accelerators to embed CNN show tremendous performance and energy efficiency, they are not suitable for SR tasks regarding off-chip memory accesses. In this work, we present eSRCNN, a framework that enables performing energy-efficient SR tasks on diverse embedded CNN accelerators by decreasing off-chip memory accesses. To reduce off-chip memory accesses, our framework consists of three steps: a network reformation using a cross-layer weight scaling, a precision minimization with priority-based quantization, and an activation map compression exploiting a data locality. As a result, the energy consumption of off-chip memory accesses is reduced up to 71.89% with less than 3.52% area overhead.
KW - Activation Map Compression
KW - Convolutional Neural Network
KW - Network Reformation
KW - Super-Resolution
UR - http://www.scopus.com/inward/record.url?scp=85077790066&partnerID=8YFLogxK
U2 - 10.1109/ICCAD45719.2019.8942086
DO - 10.1109/ICCAD45719.2019.8942086
M3 - Conference contribution
AN - SCOPUS:85077790066
T3 - IEEE/ACM International Conference on Computer-Aided Design, Digest of Technical Papers, ICCAD
BT - 2019 IEEE/ACM International Conference on Computer-Aided Design, ICCAD 2019 - Digest of Technical Papers
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 4 November 2019 through 7 November 2019
ER -