A Real-Time 17-Scale Object Detection Accelerator With Adaptive 2000-Stage Classification in 65 nm CMOS

Kim, Minkyu; Mohanty, Abinash; Kadetotad, Deepak; Wei, Luning; He, Xiaofei; Cao, Yu; Seo, Jae-sun

doi:10.1109/tcsi.2019.2921714

Cited by 6 publications

(2 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Accelerating deep neural network processing in edge computing using energy-efficient platforms is an important goal [ 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ]. Currently, most object detection and classification models are carried out in graphics processing units.…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, many lightweight approaches with low-power consumption and low-computational performance have emerged recently. A few dedicated neural network accelerators have been implemented on FPGA hardware platforms [ 12 , 14 , 17 , 21 , 23 ], while several authors proposed ASIC-based neural network accelerators [ 13 , 15 , 16 , 18 , 19 , 22 ]. Samimi et al [ 20 ] proposed a technique based on the residue number system to improve the energy efficiency of deep neural network processing.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An Approximate GEMM Unit for Energy-Efficient Object Detection

Pilipović

Risojević

Božič

et al. 2021

Sensors

View full text Add to dashboard Cite

Edge computing brings artificial intelligence algorithms and graphics processing units closer to data sources, making autonomy and energy-efficient processing vital for their design. Approximate computing has emerged as a popular strategy for energy-efficient circuit design, where the challenge is to achieve the best tradeoff between design efficiency and accuracy. The essential operation in artificial intelligence algorithms is the general matrix multiplication (GEMM) operation comprised of matrix multiplication and accumulation. This paper presents an approximate general matrix multiplication (AGEMM) unit that employs approximate multipliers to perform matrix–matrix operations on four-by-four matrices given in sixteen-bit signed fixed-point format. The synthesis of the proposed AGEMM unit to the 45 nm Nangate Open Cell Library revealed that it consumed only up to 36% of the area and 25% of the energy required by the exact general matrix multiplication unit. The AGEMM unit is ideally suited to convolutional neural networks, which can adapt to the error induced in the computation. We evaluated the AGEMM units’ usability for honeybee detection with the YOLOv4-tiny convolutional neural network. The results implied that we can deploy the AGEMM units in convolutional neural networks without noticeable performance degradation. Moreover, the AGEMM unit’s employment can lead to more area- and energy-efficient convolutional neural network processing, which in turn could prolong sensors’ and edge nodes’ autonomy.

show abstract