Sungrae Kim

Sungrae Kim

Auto-generated profile, based on publicly available information

Sungrae Kim

Department of Electrical and Information Engineering, Seoul National University of Science and Technology, Seoul, South Korea

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 January 2021 in IEEE Access
Verified
Journal Article
Open access
SKSungrae Kim
Sungrae Kim
HKHyun Kim
Hyun Kim
In the field of object detection, deep learning has greatly improved accuracy compared to previous algorithms and has been used widely in recent years. However, object detection using deep learning requires many hardware (HW) resources due to the huge computations for high performance, making it very difficult to run real-time on embedded platforms. Therefore, various compression methods have been studied to solve this problem. In particular, quantization methods greatly reduce the computational burden of deep learning by reducing the number of bits used for weights and activation functions in deep learning. However, most of these existing studies targeted only object classification and cannot be applied to object detection. Furthermore, most of the existing quantization studies are based on floating-point operations, which requires additional effort when implementing HW accelerators. This paper proposes an HW-friendly fixed-point-based quantization method that can also be applied to object detection. In the proposed method, the center of the weight distribution is adjusted to zero by subtracting the mean of weight parameters before quantization, and the retraining process is iteratively applied to minimize the accuracy drop caused by quantization. Furthermore, while applying the proposed method to object detection, performance degradation is minimized by considering the minimum and maximum values of weight parameters of deep learning networks. When applying the proposed quantization method to representative one-stage object detectors, You Only Look Once v3 and v4 (YOLOv3 and YOLOv4), detection accuracy similar to the original networks (i.e., YOLOv3 and YOLOv4) with a single-precision floating-point format (32-bit) is maintained despite expressing weights with only about 20% of the bits compared to a single-precision floating-point format in COCO dataset.

29 September 2016 in Modular and Offsite Construction (MOC) Summit Proceedings
Verified
Journal Article
Open access
PDF file
Youngsoo JungYoungsoo Jung
Youngsoo Jung
ILIkhaeng Lee
Ikhaeng Lee
SKSungrae Kim
Sungrae Kim
YLYunsub Lee
Yunsub Lee
In an attempt to disseminate traditional Korean culture, a comprehensive research has been initiated by Korean government in order to modernize Korean traditional housing (Hanok) with the objectives of improved energy performance and affordable construction cost. This äóÖHanokäó» project encompasses a wide spectrum of housing research including public policies, planning methods, standard design documents, new building materials and methods, construction standards, maintenance manuals, and advanced IT applications in an integrated manner. One of the biggest challenges in this äóÖHanokäó» project was to modernize the äóÖtraditional timber structureäó» for industrialization, while keeping the traditional way of aesthetic representation. As a solution to meet this complicated requirement, a hybrid timber system was developed by combining traditional methods and industrialized modular members. Different timbers and methods are used together for different part of house elements resulting in cost reduction by 50% for the timber frame. Major criteria for applying different methods include the aesthetic representation, economy, and deformation behaviour of wood. Automated computer numeric control (CNC) machine, standard 3D-CAD objects, standard classifications, and computer applications were also developed in order to make this hybrid system economically feasible. This paper introduces the hybrid timber system for Hanok along with supportive application systems for industrialization. Three mock-up projects, actually built as part of this research project, are compared and analysed in order to illustrate how the proposed hybrid timber system has evolved during the research and development. Lessons learned and future directions will be also briefly discussed.