• Electronics and Telecommunications Trends
• /
• v.34 no.2
• /
• pp.40-50
• /
• 2019

Considerable accuracy improvements in deep learning have recently been achieved in many applications that require large amounts of computation and expensive memory. However, recent advanced techniques for compacting and accelerating the deep learning model have been developed for deployment in lightweight devices with constrained resources. Lightweight deep learning techniques can be categorized into two schemes: lightweight deep learning algorithms (model simplification and efficient convolutional filters) in nature and transferring models into compact/small ones (model compression and knowledge distillation). In this report, we briefly summarize various lightweight deep learning techniques and possible research directions.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.3
• /
• pp.346-354
• /
• 2024

Deep learning models require high computing power due to a substantial amount of computation. It is difficult to use them in devices with limited computing environments, such as coastal surveillance equipments. In this study, a lightweight model is constructed by analyzing the weight changes of the convolutional layers during the training process based on MobileNet and then pruning the layers that affects the model less. The performance comparison results show that the lightweight model maintains performance while reducing computational load, parameters, model size, and data processing speed. As a result of this study, an effective pruning method for constructing lightweight deep learning models and the possibility of using equipment resources efficiently through lightweight models in limited computing environments such as coastal surveillance equipments are presented.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.17-25
• /
• 2023

In this paper, we propose a lightweight deep learning architecture tailored for efficient text detection in fashion design sketch images. Given the increasing prominence of Digital Transformation in the fashion industry, there is a growing emphasis on harnessing digital tools for creating fashion design sketches. As digitization becomes more pervasive in the fashion design process, the initial stages of text detection and recognition take on pivotal roles. In this study, a lightweight network was designed by building upon existing text detection deep learning models, taking into consideration the unique characteristics of apparel design drawings. Additionally, a separately collected dataset of apparel design drawings was added to train the deep learning model. Experimental results underscore the superior performance of our proposed deep learning model, outperforming existing text detection models by approximately 20% when applied to fashion design sketch images. As a result, this paper is expected to contribute to the Digital Transformation in the field of clothing design by means of research on optimizing deep learning models and detecting specialized text information.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.2
• /
• pp.77-83
• /
• 2022

In this paper, the two-stage object detection approach is proposed to implement a deep learning-based label inspection system on edge computing environments. Since the label printed on the products during the production process contains important information related to the product, it is significantly to check the label information is correct. The proposed system uses the lightweight deep learning model that able to employ in the low-performance edge computing devices, and the two-stage object detection approach is applied to compensate for the low accuracy relatively. The proposed Two-Stage object detection approach consists of two object detection networks, Label Area Detection Network and Character Detection Network. Label Area Detection Network finds the label area in the product image, and Character Detection Network detects the words in the label area. Using this approach, we can detect characters precise even with a lightweight deep learning models. The SF-YOLO model applied in the proposed system is the YOLO-based lightweight object detection network designed for edge computing devices. This model showed up to 2 times faster processing time and a considerable improvement in accuracy, compared to other YOLO-based lightweight models such as YOLOv3-tiny and YOLOv4-tiny. Also since the amount of computation is low, it can be easily applied in edge computing environments.

• Computers and Concrete
• /
• v.32 no.3
• /
• pp.327-337
• /
• 2023

Concrete is the most widely used building material, with various types including high- and ultra-high-strength, reinforced, normal, and lightweight concretes. However, accurately predicting concrete properties is challenging due to the geotechnical design code's requirement for specific characteristics. To overcome this issue, researchers have turned to new technologies like machine learning to develop proper methodologies for concrete specification. In this study, we propose a highly accurate deep learning-based predictive model to investigate the compressive strength (UCS) of lightweight concrete with natural aggregates (pumice). Our model was implemented on a database containing 249 experimental records and revealed that water, cement, water-cement ratio, fine-coarse aggregate, aggregate substitution rate, fine aggregate replacement, and superplasticizer are the most influential covariates on UCS. To validate our model, we trained and tested it on random subsets of the database, and its performance was evaluated using a confusion matrix and receiver operating characteristic (ROC) overall accuracy. The proposed model was compared with widely known machine learning methods such as MLP, SVM, and DT classifiers to assess its capability. In addition, the model was tested on 25 laboratory UCS tests to evaluate its predictability. Our findings showed that the proposed model achieved the highest accuracy (accuracy=0.97, precision=0.97) and the lowest error rate with a high learning rate (R2=0.914), as confirmed by ROC (AUC=0.971), which is higher than other classifiers. Therefore, the proposed method demonstrates a high level of performance and capability for UCS predictions.

• Journal of the Korean Society of Industry Convergence
• /
• v.25 no.6_3
• /
• pp.1285-1291
• /
• 2022

In this paper, we propose a low-cost, low-power embedded environment-based deep learning lightweight model for input images to recognize laundry management codes. Laundry franchise companies mainly use barcode recognition-based systems to record laundry consignee information and laundry information for laundry collection management. Conventional laundry collection management systems using barcodes require barcode printing costs, and due to barcode damage and contamination, it is necessary to improve the cost of reprinting the barcode book in its entirety of 1 billion won annually. It is also difficult to do. Recognition performance is improved by applying the VGG model with 7 layers, which is a reduced-transformation of the VGGNet model for number recognition. As a result of the numerical recognition experiment of service parts drawings, the proposed method obtained a significantly improved result over the conventional method with an F1-Score of 0.95.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.1-8
• /
• 2024

On-device AI technology, which can operate AI models at the edge devices to support real-time processing and privacy enhancement, is attracting attention. As intelligent IoT is applied to various industries, services utilizing the on-device AI technology are increasing significantly. However, general deep learning models require a lot of computational resources for inference and learning. Therefore, various lightweighting methods such as quantization and pruning have been suggested to operate deep learning models in embedded edge devices. Among the lightweighting methods, we analyze how to lightweight and apply deep learning models to edge computing devices, focusing on pruning technology in this paper. In particular, we utilize dynamic and static pruning techniques to evaluate the inference speed, accuracy, and memory usage of a lightweight AI vision model. The content analyzed in this paper can be used for intelligent video control systems or video security systems in autonomous vehicles, where real-time processing are highly required. In addition, it is expected that the content can be used more effectively in various IoT services and industries.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.6
• /
• pp.319-325
• /
• 2023

In this study, we proposed a method for hyperparameter optimization in the building and training of a deep learning model designed to process point cloud data collected by a millimeter-wave radar system. The primary aim of this study is to facilitate the deployment of a baseline model in resource-constrained IoT devices. We evaluated a RadHAR baseline deep learning model trained on a public dataset composed of point clouds representing five distinct human activities. Additionally, we introduced a coarse-to-fine hyperparameter optimization procedure, showing substantial potential to enhance model efficiency without compromising predictive performance. Experimental results show the feasibility of significantly reducing model size without adversely impacting performance. Specifically, the optimized model demonstrated a 3.3% improvement in classification accuracy despite a 16.8% reduction in number of parameters compared th the baseline model. In conclusion, this research offers valuable insights for the development of deep learning models for resource-constrained IoT devices, underscoring the potential of hyperparameter optimization and model size reduction strategies. This work contributes to enhancing the practicality and usability of deep learning models in real-world environments, where high levels of accuracy and efficiency in data processing and classification tasks are required.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.2
• /
• pp.51-58
• /
• 2023

This paper proposes a deep learning method for estimating the heart rate from facial videos. Our proposed method estimates remote photoplethysmography (rPPG) signals to predict the heart rate. Although there have been proposed several methods for estimating rPPG signals, most previous methods can not be utilized in low-power single board computers due to their computational complexity. To address this problem, we construct a lightweight student model and employ a knowledge distillation technique to reduce the performance degradation of a deeper network model. The teacher model consists of 795k parameters, whereas the student model only contains 24k parameters, and therefore, the inference time was reduced with the factor of 10. By distilling the knowledge of the intermediate feature maps of the teacher model, we improved the accuracy of the student model for estimating the heart rate. Experiments were conducted on the UBFC-rPPG dataset to demonstrate the effectiveness of the proposed method. Moreover, we collected our own dataset to verify the accuracy and processing time of the proposed method on a real-world dataset. Experimental results on a NVIDIA Jetson Nano board demonstrate that our proposed method can infer the heart rate in real time with the mean absolute error of 2.5183 bpm.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.9
• /
• pp.2505-2528
• /
• 2023

The field of facial expression recognition (FER) has been actively researched to improve human-computer interaction. In recent years, deep learning techniques have gained popularity for addressing FER, with numerous studies proposing end-to-end frameworks that stack or widen significant convolutional neural network layers. While this has led to improved performance, it has also resulted in larger model sizes and longer inference times. To overcome this challenge, our work introduces a novel lightweight model architecture. The architecture incorporates three key factors: Depth-wise Separable Convolution, Residual Block, and Attention Modules. By doing so, we aim to strike a balance between model size, inference speed, and accuracy in FER tasks. Through extensive experimentation on popular benchmark FER datasets, our proposed method has demonstrated promising results. Notably, it stands out due to its substantial reduction in parameter count and faster inference time, while maintaining accuracy levels comparable to other lightweight models discussed in the existing literature.