• Journal of Korea Multimedia Society
• /
• v.25 no.1
• /
• pp.1-9
• /
• 2022

Ship detection in synthetic aperture radar (SAR) images is an important application in marine monitoring for the military and civilian domains. Over the past decade, object detection has achieved significant progress with the development of convolutional neural networks (CNNs) and lot of labeled databases. However, due to difficulty in collecting and labeling SAR images, it is still a challenging task to solve SAR ship detection CNNs. To overcome the problem, some methods have employed conventional data augmentation techniques such as flipping, cropping, and affine transformation, but it is insufficient to achieve robust performance to handle a wide variety of types of ships. In this paper, we present a novel and effective approach for deep SAR ship detection, that exploits label-rich Electro-Optical (EO) images. The proposed method consists of two components: a data augmentation network and a ship detection network. First, we train the data augmentation network based on conditional generative adversarial network (cGAN), which aims to generate additional SAR images from EO images. Since it is trained using unpaired EO and SAR images, we impose the cycle-consistency loss to preserve the structural information while translating the characteristics of the images. After training the data augmentation network, we leverage the augmented dataset constituted with real and translated SAR images to train the ship detection network. The experimental results include qualitative evaluation of the translated SAR images and the comparison of detection performance of the networks, trained with non-augmented and augmented dataset, which demonstrates the effectiveness of the proposed framework.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.607-610
• /
• 2021

When learning a model in supervised learning, input data and the label of the data are required. However, labeling is high cost task and if automated, there is no guarantee that the label will always be correct. In the case of supervised learning in such a noisy labels environment, the accuracy of the model increases at the initial stage of learning, but decrease significantly after a certain period of time. There are various methods to solve the noisy label problem. But in most cases, the probability predicted by the model is used as the pseudo label. So, we proposed a method to predict the true label more quickly by refining the probabilities predicted by the model. Result of experiments on the same environment and dataset, it was confirmed that the performance improved and converged faster. Through this, it can be applied to methods that use the probability distribution predicted by the model among existing studies. And it is possible to reduce the time required for learning because it can converge faster in the same environment.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.2
• /
• pp.561-568
• /
• 2023

Various defects occur in concrete structures due to internal and external environments. If there is a defect, it is important to efficiently identify and maintain it because there is a problem with the structural safety of concrete. However, recent deep learning research has focused on cracks in concrete, and studies on exfoliation and contamination are lacking. In this study, focusing on exfoliation and contamination, which are difficult to label, four models were developed and their performance evaluated through unlabelling method, filtering method, the convergence of transfer learning based k-means clustering. As a result of the analysis, the convergence model classified the defects in the most detail and could increase the efficiency compared to direct labeling. It is hoped that the results of this study will contribute to the development of deep learning models for various types of defects that are difficult to label in the future.

• Communications of Mathematical Education
• /
• v.34 no.3
• /
• pp.277-297
• /
• 2020

The purpose of this study was to identify and compare the mathematics educational values of pre-service and in-service elementary school teachers. For this purpose, we implemented a questionnaire investigating mathematics educational values and used principal component analysis which resulted in six components. These components were named as fun, problem-solving, representation, computation, ability, and explanation through systematic labeling processes. Both pre-service and in-service elementary school teachers considered problem-solving the most important and there was no statistical difference between the teacher groups. They also considered fun the least important and in-service elementary school teachers regarded it more important than pre-service counterparts did. All value components except explanation were regarded as important by in-service elementary school teachers, fourth-year pre-service teachers, and first-year pre-service teachers in order. The result of noticeable differences between pre-service and in-service elementary school teachers implies that actual teaching experience may affect teachers' mathematics educational values more than teacher preparation programs. Based on these findings, we need to discuss what should be regarded as important and worthwhile in teacher preparation programs to establish mathematics educational values for pre-service teachers. We also need to confirm whether the mathematics educational values by in-service elementary school teachers may be in line with what has been pursued in the national mathematics curriculum.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.43 no.6 s.312
• /
• pp.28-35
• /
• 2006

Statistical models of shape variability based on active shape models (ASMs) have been successfully utilized to perform segmentation and recognition tasks in two-dimensional (2D) images. Three-dimensional (3D) model-based approaches are more promising than 2D approaches since they can bring in more realistic shape constraints for recognizing and delineating the object boundary. For 3D model-based approaches, however, building the 3D shape model from a training set of segmented instances of an object is a major challenge and currently it remains an open problem in building the 3D shape model, one essential step is to generate a point distribution model (PDM). Corresponding landmarks must be selected in all1 training shapes for generating PDM, and manual determination of landmark correspondences is very time-consuming, tedious, and error-prone. In this paper, we propose a novel automatic method for generating 3D statistical shape models. Given a set of training 3D shapes, we generate a 3D model by 1) building the mean shape fro]n the distance transform of the training shapes, 2) utilizing a tetrahedron method for automatically selecting landmarks on the mean shape, and 3) subsequently propagating these landmarks to each training shape via a distance labeling method. In this paper, we investigate the accuracy and compactness of the 3D model for the human liver built from 50 segmented individual CT data sets. The proposed method is very general without such assumptions and can be applied to other data sets.

• Journal of Korea Multimedia Society
• /
• v.8 no.11
• /
• pp.1496-1509
• /
• 2005

In this paper, an adaptive license plate recognition system which detects and recognizes license plate at real-time by using projected plane convolution and Decision Tree Classifier is proposed. And it was tested in circumstances which presence of complex background. Generally, in expressway tollgate or gateway of parking lots, it is very difficult to detect and segment license plate because of size, entry angle and noisy problem of vehicles due to CCD camera and road environment. In the proposed algorithm, we suggested to extract license plate candidate region after going through image acquisition process with inputted real-time image, and then to compensate license size as well as gradient of vehicle with change of vehicle entry position. The proposed algorithm can exactly detect license plate using accumulated edge, projected convolution and chain code labeling method. And it also segments letter of license plate using adaptive binary method. And then, it recognizes license plate letter by applying hybrid pattern vector method. Experimental results show that the proposed algorithm can recognize the front and rear direction license plate at real-time in the presence of complex background environments. Accordingly license plate detection rate displayed $98.8\%$ and $96.5\%$ successive rate respectively. And also, from the segmented letters, it shows $97.3\%$ and $96\%$ successive recognition rate respectively.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.10
• /
• pp.707-716
• /
• 2017

The advances in image display devices necessitate display images suitable for the user's purpose. The display devices should be able to provide object-based image information when a depthmap is required. In this paper, we represent the algorithm using a histogram-based image segmentation method for depthmap generation. In the conventional K-means clustering algorithm, the number of centroids is parameterized, so existing K-means algorithms cannot adaptively determine the number of clusters. Further, the problem of K-means algorithm tends to sink into the local minima, which causes over-segmentation. On the other hand, the proposed algorithm is adaptively able to select centroids and can stand on the basis of the histogram-based algorithm considering the amount of computational complexity. It is designed to show object-based results by preventing the existing algorithm from falling into the local minimum point. Finally, we remove the over-segmentation components through connected-component labeling algorithm. The results of proposed algorithm show object-based results and better segmentation results of 0.017 and 0.051, compared to the benchmark method in terms of Probabilistic Rand Index(PRI) and Segmentation Covering(SC), respectively.

• Korean Journal of Remote Sensing
• /
• v.15 no.2
• /
• pp.147-158
• /
• 1999

Remote sensing technique has offered better understanding of our environment for the decades by providing useful level of information on the landcover. In many applications using the remotely sensed data, digital image processing methodology has been usefully employed to characterize the features in the data and develop the models. Random field models, especially Markov Random Field (MRF) models exploiting spatial relationships, are successfully utilized in many problems such as texture modeling, region labeling and so on. Usually, remotely sensed imagery are very large in nature and the data increase greatly in the problem requiring temporal data over time period. The time required to process increasing larger images is not linear. In this study, the methodology to reduce the computational cost is investigated in the utilization of the Markov Random Field. For this, multiresolution framework is explored which provides convenient and efficient structures for the transition between the local and global features. The computational requirements for parameter estimation of the MRF model also become excessive as image size increases. A Bayesian approach is investigated as an alternative estimation method to reduce the computational burden in estimation of the parameters of large images.

• Journal of Advanced Navigation Technology
• /
• v.25 no.1
• /
• pp.115-123
• /
• 2021

Pine nuts are Korea's representative nut forest products and profitable crops. However, pine nuts are harvested by climbing the trees themselves, thus the risk is high. In order to solve this problem, it is necessary to harvest pine nuts using a robot or an unmanned aerial vehicle(UAV). In this paper, we propose a deep learning based detection method for harvesting pine nut in UAV aerial images. For this, a video was recorded in a real pine forest using UAV, and a data augmentation technique was used to supplement a small number of data. As the data for 3D detection, Unity3D was used to model the virtual pine nut and the virtual environment, and the labeling was acquired using the 3D transformation method of the coordinate system. Deep learning algorithms for detection of pine nuts distribution area and 2D and 3D detection of pine nuts objects were used DeepLabV3+, YOLOv4, and CenterNet, respectively. As a result of the experiment, the detection rate of pine nuts distribution area was 82.15%, the 2D detection rate was 86.93%, and the 3D detection rate was 59.45%.

• Korean Journal of Remote Sensing
• /
• v.38 no.6_2
• /
• pp.1633-1641
• /
• 2022

Urban population concentration and indiscriminate development are causing various environmental problems such as air pollution and heat island phenomena, and causing human resources to deteriorate the damage caused by natural disasters. Urban trees have been proposed as a solution to these urban problems, and actually play an important role, such as providing environmental improvement functions. Accordingly, quantitative measurement and analysis of individual trees in urban trees are required to understand the effect of trees on the urban environment. However, the complexity and diversity of urban trees have a problem of lowering the accuracy of single tree detection. Therefore, we conducted a study to effectively detect trees in Dongjak-gu using high-resolution aerial images that enable effective detection of tree objects and You Only Look Once Version 5 (YOLOv5), which showed excellent performance in object detection. Labeling guidelines for the construction of tree AI learning datasets were generated, and box annotation was performed on Dongjak-gu trees based on this. We tested various scale YOLOv5 models from the constructed dataset and adopted the optimal model to perform more efficient urban tree detection, resulting in significant results of mean Average Precision (mAP) 0.663.