An Ultra-Low-Power Analog-Digital Hybrid CNN Face Recognition Processor Integrated with a CIS for Always-on Mobile Devices

Kim, Jihoon; Kim, Changhyeon; Kim, Kwantae; Yoo, Hoi‐Jun

doi:10.1109/iscas.2019.8702698

Cited by 27 publications

(29 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The FD in the proposed CIS is a binary classification, which assesses whether there is a face or not on the image frames. It should be noted that the proposed CIS supports FD-only, unlike previous works in [12,13] that support FD and FR. Since the proposed a-LWCNN circuits can be implemented in column-parallel circuits, which are correlated double sampling (CDS), pixel data of every row can be read out to perform convolution functions simultaneously, thereby improving processing speed and enabling real-time operation (with a maximum frame rate of 120 fps).…”

Section: Introductionmentioning

confidence: 90%

“…In order to enhance the power efficiency, either the parts of the CNN circuits or that of the CVP can be implemented with the CIS in a chip, as shown in Figure 1b. In this case, since the CIS and CNN can be performed in a single chip, redundant and power-hungry blocks like ADC can be eliminated to further reduce power consumption [13]. Figure 2a,b shows the different types of low-power face detection (FD) and face recognition (FR) system architecture previously proposed for user-authentication applications [12,13].…”

Section: Introductionmentioning

confidence: 99%

“…In this case, since the CIS and CNN can be performed in a single chip, redundant and power-hungry blocks like ADC can be eliminated to further reduce power consumption [13]. Figure 2a,b shows the different types of low-power face detection (FD) and face recognition (FR) system architecture previously proposed for user-authentication applications [12,13]. For the FD operation, Figure 2a uses analog Haar-like filter circuit (AHFC) while Figure 2b uses analog CNN circuits.…”

Section: Introductionmentioning

confidence: 99%

“…The contents of the paper are as follows: Section 2 discusses the . Different types of face detection (FD) system architectures for (a) Haar-like FD [12], (b) analog-digital hybrid CNN FD [13], and (c) the proposed analog lightweight CNN (a-LWCNN) FD.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Design of an Always-On Image Sensor Using an Analog Lightweight Convolutional Neural Network

Choi

Lee

Son

et al. 2020

Sensors

View full text Add to dashboard Cite

This paper presents an always-on Complementary Metal Oxide Semiconductor (CMOS) image sensor (CIS) using an analog convolutional neural network for image classification in mobile applications. To reduce the power consumption as well as the overall processing time, we propose analog convolution circuits for computing convolution, max-pooling, and correlated double sampling operations without operational transconductance amplifiers. In addition, we used the voltage-mode MAX circuit for max pooling in the analog domain. After the analog convolution processing, the image data were reduced by 99.58% and were converted to digital with a 4-bit single-slope analog-to-digital converter. After the conversion, images were classified by the fully connected processor, which is traditionally performed in the digital domain. The measurement results show that we achieved an 89.33% image classification accuracy. The prototype CIS was fabricated in a 0.11 μm 1-poly 4-metal CIS process with a standard 4T-active pixel sensor. The image resolution was 160 × 120, and the total power consumption of the proposed CIS was 1.12 mW with a 3.3 V supply voltage and a maximum frame rate of 120.

show abstract

Section: Introductionmentioning

confidence: 90%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Design of an Always-On Image Sensor Using an Analog Lightweight Convolutional Neural Network

Choi

Lee

Son

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…ASPs are also widely used in biopotential acquisition and analysis systems, finding recent applications in the current-mode front-end proposed in [21] and the electrocardiogram feature detector proposed in [22]. There have also been recent demonstrations of ASPs in machine learning, such as the convolutional neural network processor shown in [23] and the framework for computing Gaussian kernels shown in [24].…”

Section: Proposal Novelty and Overviewmentioning

confidence: 99%

An Acoustic Vehicle Detector and Classifier Using a Reconfigurable Analog/Mixed-Signal Platform

Bhattacharyya

Andryzcik

Graham

2020

JLPEA

View full text Add to dashboard Cite

The wireless sensor nodes used in a growing number of remote sensing applications are deployed in inaccessible locations or are subjected to severe energy constraints. Audio-based sensing offers flexibility in node placement and is popular in low-power schemes. Thus, in this paper, a node architecture with low power consumption and in-the-field reconfigurability is evaluated in the context of an acoustic vehicle detection and classification (hereafter “AVDC”) scenario. The proposed architecture utilizes an always-on field-programmable analog array (FPAA) as a low-power event detector to selectively wake a microcontroller unit (MCU) when a significant event is detected. When awoken, the MCU verifies the vehicle class asserted by the FPAA and transmits the relevant information. The AVDC system is trained by solving a classification problem using a lexicographic, nonlinear programming algorithm. On a testing dataset comprising of data from ten cars, ten trucks, and 40 s of wind noise, the AVDC system has a detection accuracy of 100%, a classification accuracy of 95%, and no false alarms. The mean power draw of the FPAA is 43 μ W and the mean power consumption of the MCU and radio during its validation and wireless transmission process is 40.9 mW. Overall, this paper demonstrates that the utilization of an FPAA-based signal preprocessor can greatly improve the flexibility and power consumption of wireless sensor nodes.

show abstract

A CMOS rectified linear unit operating in weak inversion for memristive neuromorphic circuits

Shi

2022

Integration

View full text Add to dashboard Cite

An Ultra-Low-Power Analog-Digital Hybrid CNN Face Recognition Processor Integrated with a CIS for Always-on Mobile Devices

Cited by 27 publications

References 8 publications

Design of an Always-On Image Sensor Using an Analog Lightweight Convolutional Neural Network

Design of an Always-On Image Sensor Using an Analog Lightweight Convolutional Neural Network

An Acoustic Vehicle Detector and Classifier Using a Reconfigurable Analog/Mixed-Signal Platform

A CMOS rectified linear unit operating in weak inversion for memristive neuromorphic circuits

Contact Info

Product

Resources

About