• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.10a
• /
• pp.625-627
• /
• 2018

브레이크 패드 제작 공정에서 문자 인식은 사람이 직접 인식하거나 컴퓨터 비전 기술의 역할이었다. 하지만 사람의 인식 오류나 잉크가 번진 문자같은 새로운 형태의 문자를 인식하지 못하는 비전 기술의 단점 등 많은 한계가 존재했다. 본 논문에서는 C/CUDA로 설계한 Single Shot Multibox Detector 기반 Inference Program 을 통해 더 정확한 문자인식 결과를 제시하고, CUDA를 이용한 향상된 연산속도를 통해 실시간 문자 인식이 가능하도록 하였다. 문자 인식 정확도는 약 96.6%로 기존 비전 기술보다 더 뛰어난 성능을 보였다.

• Smart Structures and Systems
• /
• v.28 no.6
• /
• pp.811-825
• /
• 2021

High-voltage isolating switches play a paramount role in ensuring the safety of power supply systems. However, their exposure to outdoor environmental conditions may cause serious physical defects, which may result in great risk to power supply systems and society. Image processing-based methods have been used for anomaly detection. However, their accuracy is affected by numerous uncertainties due to manually extracted features, which makes the anomaly detection of isolating switches still challenging. In this paper, a vision-based anomaly detection method for isolating switches, which uses the rotational angle of the switch system for more accurate and direct anomaly detection with the help of deep learning (DL) and image processing methods (Single Shot Multibox Detector (SSD), improved frame differencing method, and Hough transform), is proposed. The SSD is a deep learning method for object classification and localization. In addition, an improved frame differencing method is introduced for better feature extraction and a hough transform method is adopted for rotational angle calculation. A number of experiments are conducted for anomaly detection of single and multiple switches using video frames. The results of the experiments demonstrate that the SSD outperforms the You-Only-Look-Once network. The effectiveness and robustness of the proposed method have been proven under various conditions, such as different illumination and camera locations using 96 videos from the experiments.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.9-12
• /
• 2019

지난 수 년 동안 계속해서 일반 실상 카메라를 이용한 영상분석기술에 대한 연구가 활발히 진행되고 있다. 최근에는 딥러닝 기술을 적용한 지능형 영상분석기술로 발전해 왔으며 국방기지방호, CCTV, 사용자 얼굴인식, 머신비전, 자동차, 드론 산업이 활성화되면서 많은 시너지를 효과를 일으키고 있다. 그러나 어두운 밤과 안개, 날씨, 연기 등 다양한 여건에서 따라서 카메라의 영상분석 정확성 감소와 오류가 수반될 수 있으며 일반적으로 딥러닝 기술을 활용하기 위해서는 고사양의 GPU를 필요로 하기 때문에 다른 추가적인 시스템이 요구된다. 이에 본 연구에서는 열적외선 영상의 객체 검출에 적용하기 위해 SSD(Single Shot MultiBox Detector) 기반의 경량적인 MobilNet 네트워크로 재구성하여, 모바일 기기 등 낮은 사양의 낮은 임베디드 시스템에서도 활용 할 수 있는 방법을 제안한다. 모의 실험결과 제안된 방식의 모델은 적외선 열화상 카메라에서 객체검출과 학습시간이 줄어든 것을 확인 할 수 있었다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.1
• /
• pp.141-144
• /
• 2020

Convolutional neural networks (CNNs) show high performance in computer vision tasks including object detection, but a lot of weight storage and computation is required. In this paper, a pruning scheme is applied to CNNs for object detection, which can remove much amount of weights with a negligible performance degradation. Contrary to the previous ones, the pruning scheme applied in this paper considers the base accelerator architecture. With the consideration, the pruned CNNs can be efficiently performed on an ASIC or FPGA accelerator. Even with the constrained pruning, the resulting CNN shows a negligible degradation of detection performance, less-than-1% point degradation of mAP on VOD0712 test set. With the proposed scheme, CNNs can be applied to objection dtection efficiently.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.348-350
• /
• 2021

The core technology of augmented reality is object recognition technology. Recently, due to the development of various artificial intelligence algorithms such as CNN, it has become possible to effectively distinguish specific objects from images. It is possible to realize more realistic and immersive augmented reality contents only when technology for recognizing objects quickly and accurately is secured. In this study, an object recognition model using SSD (single shot multibox detector) and an object recognition model using YOLO were compared and evaluated.

• Journal of IKEEE
• /
• v.23 no.3
• /
• pp.777-784
• /
• 2019

A pre-fire awareness and automatic notification system are required because it is possible to minimize the damage if the fire situation is precisely detected after a fire occurs in a place where people are unusual or in a mountainous area. In this study, we developed a RaspberryPi-based fire recognition system using Faster-recurrent convolutional neural network (F-RCNN) and single shot multibox detector (SSD) and demonstrated a fire alarm system that works with power line communication. Image recognition was performed with a pie camera of RaspberryPi, and the detected fire image was transmitted to a monitoring PC through an inductive power line communication network. The frame rate per second (fps) for each learning model was 0.05 fps for Faster-RCNN and 1.4 fps for SSD. SSD was 28 times faster than F-RCNN.

• Smart Structures and Systems
• /
• v.23 no.4
• /
• pp.359-371
• /
• 2019

As part of a structural health monitoring system, the relative geometric relationship between a ship and bridge has been recognized as important for bridge authorities and ship owners to avoid ship-bridge collision. This study proposes a novel computer vision method for the real-time geometric parameter identification of moving ships based on a single shot multibox detector (SSD) by using transfer learning techniques and monocular vision. The identification framework consists of ship detection (coarse scale) and geometric parameter calculation (fine scale) modules. For the ship detection, the SSD, which is a deep learning algorithm, was employed and fine-tuned by ship image samples downloaded from the Internet to obtain the rectangle regions of interest in the coarse scale. Subsequently, for the geometric parameter calculation, an accurate ship contour is created using morphological operations within the saturation channel in hue, saturation, and value color space. Furthermore, a local coordinate system was constructed using projective geometry transformation to calculate the geometric parameters of ships, such as width, length, height, localization, and velocity. The application of the proposed method to in situ video images, obtained from cameras set on the girder of the Wuhan Yangtze River Bridge above the shipping channel, confirmed the efficiency, accuracy, and effectiveness of the proposed method.

• Journal of Broadcast Engineering
• /
• v.23 no.3
• /
• pp.361-368
• /
• 2018

As autonomous driving research is rapidly developing, pedestrian detection study is also successfully investigated. However, most of the study utilizes color image datasets and those are relatively easy to detect the pedestrian. In case of color images, the scene should be exposed by enough light in order to capture the pedestrian and it is not easy for the conventional methods to detect the pedestrian if it is the other case. Therefore, in this paper, we propose deep neural network (DNN)-based multi-spectrum pedestrian detection method using color and thermal images. Based on single-shot multibox detector (SSD), we propose fusion network structures which simultaneously employ color and thermal images. In the experiment, we used KAIST dataset. We showed that proposed SSD-H (SSD-Halfway fusion) technique shows 18.18% lower miss rate compared to the KAIST pedestrian detection baseline. In addition, the proposed method shows at least 2.1% lower miss rate compared to the conventional halfway fusion method.

• The Journal of Korean Institute of Information Technology
• /
• v.17 no.5
• /
• pp.13-20
• /
• 2019

Recently, object detection technology based on CNN has been actively studied. Object detection technology is used as an important technology in autonomous vehicles, intelligent image analysis, and so on. In this paper, we propose a rotation change robust object detection system by applying MI-FL (Moment Invariant-Feature Layer) to SSD (Single Shot Multibox Detector) which is one of CNN-based object detectors. First, the features of the input image are extracted based on the VGG network. Then, a total of six feature layers are applied to generate bounding boxes by predicting the location and type of object. We then use the NMS algorithm to get the bounding box that is the most likely object. Once an object bounding box has been determined, the invariant moment feature of the corresponding region is extracted using MI-FL, and stored and learned in advance. In the detection process, it is possible to detect the rotated image more robust than the conventional method by using the previously stored moment invariant feature information. The performance improvement of about 4 ~ 5% was confirmed by comparing SSD with existing SSD and MI-FL.

• Journal of Korea Multimedia Society
• /
• v.22 no.4
• /
• pp.424-431
• /
• 2019

In this paper, we propose a real-time hand gesture recognition algorithm to eliminate the inconvenience of using hand controllers in VR applications. The user's 3D hand coordinate information is detected by leap motion sensor and then the coordinates are generated into two dimensional image. We classify hand gestures in real-time by learning the imaged 3D hand coordinate information through SSD(Single Shot multibox Detector) model which is one of CNN(Convolutional Neural Networks) models. We propose to use all 3 channels rather than only one channel. A sliding window technique is also proposed to recognize the gesture in real time when the user actually makes a gesture. An experiment was conducted to measure the recognition rate and learning performance of the proposed model. Our proposed model showed 99.88% recognition accuracy and showed higher usability than the existing algorithm.