• 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 Korea Society of Digital Industry and Information Management
• /
• v.19 no.4
• /
• pp.127-136
• /
• 2023

Many research results on the traffic enforcement of illegal driving of twowheeled motor vehicles using license plate recognition are introduced. Deep learning convolutional neural networks can be used for character and word recognition of license plates because of better generalization capability compared to traditional Backpropagation neural networks. In the plates of twowheeled motor vehicles, the interdependent government and city words are included. If we implement the mutually independent word recognizers using error correction rules for two word recognition results, efficient license plate recognition results can be derived. The CPU based convolutional neural network without library under real time processing has an advantage of low cost real application compared to GPU based convolutional neural network with library. In this paper twowheeled motor vehicle license plate recognition algorithm is introduced using CPU based deep-learning convolutional neural network. The experimental results show that the proposed plate recognizer has 96.2% success rate for outdoor twowheeled motor vehicle images in real time.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.14 no.6
• /
• pp.1207-1214
• /
• 2019

Technologies of the 4th industrial revolution have been unknowingly seeping into our lives. Many IoT based home security systems are using the convolutional neural network(CNN) as good biometrics to recognize a face and protect home and family from intruders since CNN has demonstrated its excellent ability in image recognition. In this paper, three layouts of CNN for 2D and 2.5D image of small dataset with various input image size and filter size are explored. The simulation results show that the layout of CNN with 50*50 input size of 2.5D image, 2 convolution and max pooling layer, and 3*3 filter size for small dataset of 2.5D image is optimal for a home security system with recognition accuracy of 0.966. In addition, the longest CPU time consumption for one input image is 0.057S. The proposed layout of CNN for a face recognition is suitable to control the actuators in the home security system because a home security system requires good face recognition and short recognition time.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.355-366
• /
• 2022

This paper proposes a microcode-based neural network accelerator controller for artificial intelligence accelerators that can be reconstructed using a programmable architecture and provide the advantages of low-power and ultra-small chip size. In order for the target accelerator to support various neural network models, the neural network model can be converted into microcode through microcode compiler and mounted on accelerator to control the operators of the accelerator such as datapath and memory access. While the proposed controller and accelerator can run various CNN models, in this paper, we tested them using the YOLOv2-Tiny CNN model. Using a system clock of 200 MHz, the Controller and accelerator achieved an inference time of 137.9 ms/image for VOC 2012 dataset to detect object, 99.5ms/image for mask detection dataset to detect wearing mask. When implementing an accelerator equipped with the proposed controller as a silicon chip, the gate count is 618,388, which corresponds to 65.5% reduction in chip area compared with an accelerator employing a CPU-based controller (RISC-V).

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.10
• /
• pp.4080-4097
• /
• 2020

In recent years, convolutional neural network (CNN) has become the primary method for face detection. But its shortcomings are obvious, such as expensive calculation, heavy model, etc. This makes CNN difficult to use on the mobile devices which have limited computing and storage capabilities. Therefore, the design of lightweight CNN for face detection is becoming more and more important with the popularity of smartphones and mobile Internet. Based on the CPU real-time face detector FaceBoxes, we propose a multi-task lightweight face detector, which has low computing cost and higher detection precision. First, to improve the detection capability, the squeeze and excitation modules are used to extract attention between channels. Then, the textual and semantic information are extracted by shallow networks and deep networks respectively to get rich features. Finally, the landmark detection module is used to improve the detection performance for small faces and provide landmark data for face alignment. Experiments on AFW, FDDB, PASCAL, and WIDER FACE datasets show that our algorithm has achieved significant improvement in the mean average precision. Especially, on the WIDER FACE hard validation set, our algorithm outperforms the mean average precision of FaceBoxes by 7.2%. For VGA-resolution images, the running speed of our algorithm can reach 23FPS on a CPU device.

• Journal of IKEEE
• /
• v.24 no.3
• /
• pp.685-690
• /
• 2020

We propose an efficient convolutional neural network with much lower computational complexity and higher accuracy based on MobileNet V2 for mobile or edge devices. The proposed network consists of bottleneck layers with larger expansion factors and adjusted number of channels, and excludes a few layers, and therefore, the computational complexity is reduced by half. The performance the proposed network is verified by measuring the accuracy and execution times by CPU and GPU using ImageNet100 dataset. In addition, the execution time on GPU depends on the CNN architecture.

• Journal of Sensor Science and Technology
• /
• v.29 no.1
• /
• pp.19-26
• /
• 2020

Electrocardiogram (ECG) classification has become an essential task of modern day wearable devices, and can be used to detect cardiovascular diseases. State-of-the-art Artificial Intelligence (AI)-based ECG classifiers have been designed using various artificial neural networks (ANNs). Despite their high accuracy, ANNs require significant computational resources and power. Herein, three different ANNs have been compared: multilayer perceptron (MLP), convolutional neural network (CNN), and spiking neural network (SNN) only for the ECG classification. The ANN model has been developed in Python and Theano, trained on a central processing unit (CPU) platform, and deployed on a PYNQ-Z2 FPGA board to validate the model using a Jupyter notebook. Meanwhile, the hardware accelerator is designed with Overlay, which is a hardware library on PYNQ. For classification, the MIT-BIH dataset obtained from the Physionet library is used. The resulting ANN system can accurately classify four ECG types: normal, atrial premature contraction, left bundle branch block, and premature ventricular contraction. The performance of the ECG classifier models is evaluated based on accuracy and power. Among the three AI algorithms, the SNN requires the lowest power consumption of 0.226 W on-chip, followed by MLP (1.677 W), and CNN (2.266 W). However, the highest accuracy is achieved by the CNN (95%), followed by MLP (76%) and SNN (90%).

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.3
• /
• pp.89-100
• /
• 2023

Recently, on-device artificial intelligence (AI) solutions using mobile devices and embedded edge devices have emerged in various fields, such as computer vision, to address network traffic burdens, low-energy operations, and security problems. Although vision transformer deep learning models have outperformed conventional convolutional neural network (CNN) models in computer vision, they require more computations and parameters than CNN models. Thus, they are not directly applicable to embedded edge devices with limited hardware resources. Many researchers have proposed various model compression methods or lightweight architectures for vision transformers; however, there are only a few studies evaluating the effects of model compression techniques of vision transformers on performance. Regarding this problem, this paper presents a performance evaluation of vision transformers on embedded platforms. We investigated the behaviors of three vision transformers: DeiT, LeViT, and MobileViT. Each model performance was evaluated by accuracy and inference time on edge devices using the ImageNet dataset. We assessed the effects of the quantization method applied to the models on latency enhancement and accuracy degradation by profiling the proportion of response time occupied by major operations. In addition, we evaluated the performance of each model on GPU and EdgeTPU-based edge devices. In our experimental results, LeViT showed the best performance in CPU-based edge devices, and DeiT-small showed the highest performance improvement in GPU-based edge devices. In addition, only MobileViT models showed performance improvement on EdgeTPU. Summarizing the analysis results through profiling, the degree of performance improvement of each vision transformer model was highly dependent on the proportion of parts that could be optimized in the target edge device. In summary, to apply vision transformers to on-device AI solutions, either proper operation composition and optimizations specific to target edge devices must be considered.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.3
• /
• pp.755-778
• /
• 2024

In recent years, the number of devices being connected to the internet has grown enormously, as has the intrusive behavior in the network. Thus, it is important for intrusion detection systems to report all intrusive behavior. Using deep learning and machine learning algorithms, intrusion detection systems are able to perform well in identifying attacks. However, the concern with these deep learning algorithms is their inability to identify a suitable network based on traffic volume, which requires manual changing of hyperparameters, which consumes a lot of time and effort. So, to address this, this paper offers a solution using the extended compact genetic algorithm for the automatic tuning of the hyperparameters. The novelty in this work comes in the form of modeling the problem of identifying attacks as a multi-objective optimization problem and the usage of linkage learning for solving the optimization problem. The solution is obtained using the feature map-based Convolutional Neural Network that gets encoded into genes, and using the extended compact genetic algorithm the model is optimized for the detection accuracy and latency. The CIC-IDS-2017 and 2018 datasets are used to verify the hypothesis, and the most recent analysis yielded a substantial F1 score of 99.23%. Response time, CPU, and memory consumption evaluations are done to demonstrate the suitability of this model in a fog environment.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.