• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.479-481
• /
• 2019

본 연구에서는 이미지를 분류하는 인공 신경망 가속기를 최적화했고, 이를 구현하여 기존 인공 신경망 가속기와 성능을 비교 분석했다. FPGA(Field Programmable Fate Array) 보드를 이용하여 가속기를 구현했으며, 해당 보드의 내부 메모리인 BRAM 을 FIFO(First In First Out)구조로 설계하여 메모리 시스템을 구현했다. Approximate computing 기법을 효율적으로 적용하기 위해 FWL(Fractional Word Length)최적점을 분석했고, 이를 기반으로 인공 신경망 가속기의 부동 소수점 연산을 고정 소수점 연산으로 변환했다. 구현된 인공 신경망 가속기는 기존의 인공 신경망에 비해, 약 7.4%더 효율적인 전력소모량을 보였다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.2
• /
• pp.325-328
• /
• 2022

A convolutional neural network(CNN) is widely used in the computer vision tasks, but its computing power requirement needs a design of a special circuit. Most of the computations in a CNN can be implemented efficiently in a digital circuit, but the SoftMax layer has operations unsuitable for circuit implementation, which are exponential and logarithmic functions. This paper proposes a new method to integrate the exponential and logarithmic tables of the conventional circuits into a single table. The proposed structure accesses a look-up table (LUT) only with a few maximum values, and the LUT has the result value directly. Our proposed method significantly reduces the space complexity of the SoftMax layer circuit implementation. But our resulting circuit is comparable to the original baseline with small degradation in precision.

• Journal of IKEEE
• /
• v.21 no.4
• /
• pp.388-396
• /
• 2017

Since there are many variables such as various poses, illuminations and occlusions in a face detection problem, a high performance detection system is required. Although CNN is excellent in image classification, CNN operatioin requires high-performance hardware resources. But low cost low power environments are essential for small and mobile systems. So in this paper, the CPU-FPGA integrated system is designed based on 3-stage cascade CNN architecture using small size FPGA. Adaptive Region of Interest (ROI) is applied to reduce the number of CNN operations using face information of the previous frame. We use a Field Programmable Gate Array(FPGA) to accelerate the CNN computations. The accelerator reads multiple featuremap at once on the FPGA and performs a Multiply-Accumulate (MAC) operation in parallel for convolution operation. The system is implemented on Altera Cyclone V FPGA in which ARM Cortex A-9 and on-chip SRAM are embedded. The system runs at 30FPS with HD resolution input images. The CPU-FPGA integrated system showed 8.5 times of the power efficiency compared to systems using CPU only.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1314-1320
• /
• 2019

In this paper, we propose a new method for optimizing bit operations and applying them to DNN(Deep Neural Network) in software environment. As a method for this, we propose a packing function for bitwise optimization and a masking matrix multiplication operation for application to DNN. The packing function converts 32-bit real value to 2-bit quantization value through threshold comparison operation. When this sequence is over, four 32-bit real values are changed to one 8-bit value. The masking matrix multiplication operation consists of a special operation for multiplying the packed weight value with the normal input value. And each operation was then processed in parallel using a GPU accelerator. As a result of this experiment, memory saved about 16 times than 32-bit DNN Model. Nevertheless, the accuracy was within 1%, similar to the 32-bit model.

• Journal of Advanced Navigation Technology
• /
• v.13 no.2
• /
• pp.262-271
• /
• 2009

The purpose of this study was to propose the method to recognize gestures based on inertia sensor which recognizes the movements of the user using inertia sensor and lets the user enjoy the game by comparing the recognized movements with the pre-defined movements for the game contents production. Additionally, it was tried to provide users with various data entry methods by letting them wear small controllers using three-axis accelerator sensor. Users can proceed the game by moving according to the action list printed on the screen. They can proceed the experiential games according to the accuracy and timing of their movements. If they use multiple small wireless controllers together wearing them on the major parts of hands and feet and utilize the proposed methods, they will be more interested in the game and be absorbed in it.

• Journal of IKEEE
• /
• v.24 no.3
• /
• pp.777-782
• /
• 2020

In this paper, we propose CNN-based HMI system using Doppler radar and voice sensor, and present hardware design and implementation results. To overcome the limitation of single sensor monitoring, the proposed HMI system combines data from two sensors to improve performance. The proposed system exhibits improved performance by 3.5% and 12% compared to a single radar and voice sensor-based classifier in noisy environment. In addition, hardware to accelerate the complex computational unit of CNN is implemented and verified on the FPGA test system. As a result of performance evaluation, the proposed HMI acceleration platform can be processed with 95% reduction in computation time compared to a single software-based design.

• Journal of Advanced Navigation Technology
• /
• v.25 no.4
• /
• pp.299-304
• /
• 2021

Existing human activity recognition systems detect activities through devices such as wearable sensors and cameras. However, these methods require additional devices and costs, especially for cameras, which cause privacy issue. Using WiFi signals that are already installed can solve this problem. In this paper, we propose a CNN-based human activity recognition system using channel state information of WiFi signals, and present results of designing and implementing accelerated hardware structures. The system defined four possible behaviors during studying in indoor environments, and classified the channel state information of WiFi using convolutional neural network (CNN), showing and average accuracy of 91.86%. In addition, for acceleration, we present the results of an accelerated hardware structure design for fully connected layer with the highest computation volume on CNN classifiers. As a result of performance evaluation on FPGA device, it showed 4.28 times faster calculation time than software-based system.