• Journal of Digital Convergence
• /
• v.19 no.5
• /
• pp.307-313
• /
• 2021

In this paper, a lightweight network with fewer parameters compared to the existing object detection method is proposed. In the case of the currently used detection model, the network complexity has been greatly increased to improve accuracy. Therefore, the proposed network uses EfficientNet as a feature extraction network, and the subsequent layers are formed in a pyramid structure to utilize low-level detailed features and high-level semantic features. An attention process was applied between pyramid structures to suppress unnecessary noise for prediction. All computational processes of the network are replaced by depth-wise and point-wise convolutions to minimize the amount of computation. The proposed network was trained and evaluated using the PASCAL VOC dataset. The features fused through the experiment showed robust properties for various objects through a refinement process. Compared with the CNN-based detection model, detection accuracy is improved with a small amount of computation. It is considered necessary to adjust the anchor ratio according to the size of the object as a future study.

• Journal of the Institute of Convergence Signal Processing
• /
• v.22 no.2
• /
• pp.85-90
• /
• 2021

A ADAS(Advanced Driver Assistance System) for the safe driving is an important area in autonumous car. Specially, a ADAS software using an image sensors attached in previous car is low in building cost, and utilizes for various purpose. A algorithm for detecting the break-lamp from the tail-lamp of preceding vehicle is proposed in this paper. This method can perceive the driving condition of preceding vehicle. Proposed method uses the YOLO techinicque that has a excellent performance in object tracing from real scene, and extracts the intensity variable region of break-lamp from HSV image of detected vehicle ROI(Region Of Interest). After detecting the candidate region of break-lamp, each isolated region is labeled. The break-lamp region is detected finally by using the proposed selective-attention model that percieves the shape-similarity of labeled candidate region. In order to evaluate the performance of the preceding vehicle break-lamp detection system implemented in this paper, we applied our system to the various driving images. As a results, implemented system showed successful results.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.1
• /
• pp.44-49
• /
• 2021

Recently, due to COVID-19, studies have been popularly worked to apply neural network to mask wearing automatic detection system. For applying neural networks, the 1-stage detection or 2-stage detection methods are used, and if data are not sufficiently collected, the pretrained neural network models are studied by applying fine-tuning techniques. In this paper, the system is consisted of 2-stage detection method that contain MTCNN model for face recognition and ResNet model for mask detection. The mask detector was experimented by applying five ResNet models to improve accuracy and fps in various environments. Training data used 17,217 images that collected using web crawler, and for inference, we used 1,913 images and two one-minute videos respectively. The experiment showed a high accuracy of 96.39% for images and 92.98% for video, and the speed of inference for video was 10.78fps.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1330-1339
• /
• 2022

3D object detection generally aims to detect relatively large data such as automobiles, buses, persons, furniture, etc, so it is vulnerable to small object detection. In addition, in an environment with limited resources such as embedded devices, it is difficult to apply the model because of the huge amount of computation. In this paper, the accuracy of small object detection was improved by focusing on local features using only one layer, and the inference speed was improved through the proposed knowledge distillation method from large pre-trained network to small network and adaptive quantization method according to the parameter size. The proposed model was evaluated using SUN RGB-D Val and self-made apple tree data set. Finally, it achieved the accuracy performance of 62.04% at mAP@0.25 and 47.1% at mAP@0.5, and the inference speed was 120.5 scenes per sec, showing a fast real-time processing speed.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.7
• /
• pp.1332-1337
• /
• 2006

In this paper, a noble second stage hetero-estimator is used for positioning error detection in mobile robot. Previous methods are either expensive in the case of positioning error correction or not able to detect positioning error. To overcome the latter shortage, the positioning error detection is performed using second stage hetero-estimator in motor model of mobile robot without any additional costs. A Kalman filter in the estimator gets the residual of motor current and an adaptive self-tunning filter checks the whiteness of the residual. Some simulation results show the possibility of the proposed method.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.27 no.10
• /
• pp.60-67
• /
• 1990

This paper proposes a model-based fault detection method for linear/nonlinear system having modelling errors, nonlinearities and measurement noise. The system model is represented by the unified operator [5] in order to apply to both the continuous-time and discrete-time problems. The fault detection method suggested here accounts for the effects of noise, model mismatch and nonlinearities. Modelling errors are depicted by additive forms and the nominal model denominator is fixed via prior experiments in order to quantify the nucertainty bound on the parameter estima-tion. The least square method is used to estimate the numerator parameters of the nominal model. performance than traditional methods.

• Journal of Biomedical Engineering Research
• /
• v.15 no.4
• /
• pp.447-454
• /
• 1994

A method has been proposed for the fully automatic detection of left ventricular endocardial boundary in 2D short axis echocardlogram using geometric model. The procedure has the following three distinct stages. First, the initial center is estimated by the initial center estimation algorithm which is applied to decimated image. Second, the center estimation algorithm is applied to original image and then best-fit elliptic model estimation is processed. Third, best-fit boundary is detected by the cost function which is based on the best-fit elliptic model. The proposed method shows effective result without manual intervention by a human operator.

• Proceedings of the KSRS Conference
• /
• 2008.03a
• /
• pp.20-24
• /
• 2008

차량에 부착된 CCD 카메라를 이용하여 취득된 도로 주변의 영상에 존재하는 사람의 얼굴을 추출하여 제거하는 처리를 할 경우, 사생활 침해의 문제 없이 사용자들에게 원하는 지역의 도로영상의 제공이 가능해진다. 이 실험의 목적은 차량에서 취득된 도로 주변의 칼라 영상에서 사람의 얼굴을 자동으로 추출하는 기술을 개발하는데에 있다. 도로 주변의 CCD영상에서의 얼굴 추출을 위해, HSI(색상, 채도, 명도) 칼라 모델과 YCrCb 칼라 모델을 사용하여 이들 모델에 임계값을 적용하여 피부색을 검출하였으며, 두 개의 모델을 사용한 결과 효과적인 피부색의 검출이 가능함을 확인할 수 있었다. 검출된 피부색 영역을 연결성과 밝기 차이를 이용하여 클러스터링을 실행하고 이렇게 나뉘어진 각각의 구역들에 구역의 면적, 구역내 존재하는 화소의 개수, 구역의 가로와 세로 비율 그리고 타원조건을 적용하여 얼굴 후보 구역을 결정하였다. 그리고 최종적으로 남겨진 구역을 이진화 하고, 이진화 된 영상 중 검은 부분이 5% 이상일 때 이들을 눈, 코, 입 등으로 간주하여 최종적인 얼굴로 결정하였다. 실험 결과 추출되지 않은 얼굴과 잘못 추출된 구역이 발생했으나, 얼굴에 해당하는 임계값등의 조건을 약화시킬 경우 대부분의 얼굴의 추출이 가능할 것으로 여겨지며, 추출된 구역을 흐리게 처리할 경우 오인식된 부분에 대한 사용자의 거부감도 줄일 수 있을 것 으로 예상된다.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.3
• /
• pp.109-114
• /
• 2021

GAN(Generative Adversarial Network) is an image generation model, which is composed of a generator network and a discriminator network, and generates an image similar to a real image. Since the image generated by the GAN should be similar to the actual image, a loss function is used to minimize the loss error of the generated image. However, there is a problem that the loss function of GAN degrades the quality of the image by making the learning to generate the image unstable. To solve this problem, this paper analyzes GAN-related studies and proposes an edge GAN(eGAN) using edge detection. As a result of the experiment, the eGAN model has improved performance over the existing GAN model.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2013.06a
• /
• pp.331-334
• /
• 2013

본 논문에서는 차량에 부착된 4대의 어안렌즈 카메라 영상을 이용하여 차량 주위 전방향의 주변 정보를 포함하는 정합 영상을 생성하고, 생성된 정합 영상에서 차선을 검출하는 알고리즘을 제안한다. 기존의 전방 카메라만을 이용하여 차선을 검출하는 방법들은 안개와 같이 기상 환경이 안 좋은 경우 가시거리가 짧아져 정상적인 차선 검출이 어려운 문제가 있다. 이에 반해 4대의 어안렌즈 카메라로 차량의 주변을 촬영한 영상은 기상 환경에 영향을 적게 받아 안정적인 차선 검출에 용이하다. 어안렌즈 카메라로 촬영한 영상은 왜곡이 심하기 때문에 왜곡 보정을 수행한 후 차량 위에서 아래로 내려다본 시점으로 투영 변환하여 하나의 영상으로 정합한다. 정합영상에서 관심영역을 설정한 후 차선 후보 영역을 검출하고, 검출된 후보 영역들로 차선을 직선으로 모델링한다. 점선 차선 구간이나 차량 흔들림에 대응하기 위해 직선으로 모델링된 차선 정보의 차선 각도와 차량으로부터 거리 정보를 칼만 필터 기반 추적 및 보정하여 안정적으로 차선 검출을 수행한다. 실험 결과 제안하는 방법은 실선구간에서 99.57%, 점선구간에서는 90.48%의 검출 정확도를 가진다.