• Journal of Information Processing Systems
• /
• v.18 no.2
• /
• pp.223-234
• /
• 2022

The evaluation of the print quality of 3D printing has traditionally relied on manual work using dimensional measurements. However, the dimensional measurement method has an error value that depends on the person who measures it. Therefore, we propose the design of a new print quality measurement method that can be automatically measured using the field-of-view (FOV) model and the intersection over union (IoU) technique. First, the height information of the modeling is acquired from a camera; the output is measured by a sensor; and the images of the top and isometric views are acquired from the FOV model. The height information calculates the height ratio by calculating the percentage of modeling and output, and compares the 2D contour of the object on the image using the FOV model. The contour of the object is obtained from the image for 2D contour comparison and the IoU is calculated by comparing the areas of the contour regions. The accuracy of the automated measurement technique for determining, which derives the print quality value was calculated by averaging the IoU value corrected by the measurement error and the height ratio value.

• Journal of IKEEE
• /
• v.25 no.4
• /
• pp.619-624
• /
• 2021

Vehicle orientation detection is a challenging task because the orientations of vehicles can vary in a wide range in captured images. The existing methods for oriented vehicle detection require too much computation time to be applied to a real-time system. We propose Rotate YOLO, which has a set of anchor boxes with multiple scales, ratios, and angles to predict bounding boxes. For estimating the orientation angle, we applied angle-related IoU with CIoU loss to solve the underivable problem from the calculation of SkewIoU. Evaluation results on three public datasets DLR Munich, VEDAI and UCAS-AOD demonstrate the efficiency of our approach.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.29-39
• /
• 2022

Crack detection plays an important role in the maintenance and protection of steel box girder of bridges. However, since the cracks only occupy an extremely small region of the high-resolution images captured from actual conditions, the existing methods cannot deal with this kind of image effectively. To solve this problem, this paper proposed a novel three-stage method based on deep learning technology and morphology operations. The training set and test set used in this paper are composed of 360 images (4928 × 3264 pixels) in steel girder box. The first stage of the proposed model converted high-resolution images into sub-images by using patch-based method and located the region of cracks by CBAM ResNet-50 model. The Recall reaches 0.95 on the test set. The second stage of our method uses the Attention U-Net model to get the accurate geometric edges of cracks based on results in the first stage. The IoU of the segmentation model implemented in this stage attains 0.48. In the third stage of the model, we remove the wrong-predicted isolated points in the predicted results through dilate operation and outlier elimination algorithm. The IoU of test set ascends to 0.70 after this stage. Ablation experiments are conducted to optimize the parameters and further promote the accuracy of the proposed method. The result shows that: (1) the best patch size of sub-images is 1024 × 1024. (2) the CBAM ResNet-50 and the Attention U-Net achieved the best results in the first and the second stage, respectively. (3) Pre-training the model of the first two stages can improve the IoU by 2.9%. In general, our method is of great significance for crack detection.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_3
• /
• pp.1095-1107
• /
• 2020

Flood monitoring using satellite data has been constrained by obtaining satellite images for flood peak and accurately extracting flooded areas from satellite data. Deep learning is a promising method for satellite image classification, yet the potential of deep learning-based flooded area extraction using SAR data remained uncertain, which has advantages in obtaining data, comparing to optical satellite data. This research explores the performance of SegNet and U-Net on image segmentation by extracting flooded areas in the Khorat basin, Mekong river basin, and Cagayan river basin in Thailand, Laos, and the Philippines from Sentinel-1 A/B satellite data. Results show that Global Accuracy, Mean IoU, and Mean BF Score of SegNet are 0.9847, 0.6016, and 0.6467 respectively, whereas those of U-Net are 0.9937, 0.7022, 0.7125. Visual interpretation shows that the classification accuracy of U-Net is higher than SegNet, but overall processing time of SegNet is around three times faster than that of U-Net. It is anticipated that the results of this research could be used when developing deep learning-based flood monitoring models and presenting fully automated flooded area extraction models.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.3
• /
• pp.3-12
• /
• 2015

Recently, various studies have been attempted to provide a biological information monitoring service through integrating with the web service. The medical information transmission standard ISO/IEEE 11073 PHD defines the optimized exchange protocol ISO/IEEE 11073-20601 based on the No-IP to exchange the biometric information between the ISO/IEEE 11073 agent and the manager. It's system structure based on the No-IP using ISO/IEEE 11073-20601 is not suitable for providing a remote biological information monitoring services. That is because it is difficult to provide to control and manage the biological information measurement devices, which have installed IP protocol stack at the remote. Furthermore, ACSE and CMDISE in ISO/IEEE 11073-20601 are not suitable to provide U-healthcare services based on IoT because they are complicated and difficult to implement it caused by the structural complexity. In order to solve the problems, in this paper, we propose the biological information monitoring architecture based on ISO/IEEE 11073 DIM/REST of IoT environment to provide the biological information monitoring service based on IoT. To do this, we designed biological information monitoring system architecture based on IoT and the message exchange protocol of ISO/IEEE 11073 DIM/REST between the ISO/IEEE 11073 agent and the ISO/IEEE 11073 manager. In order to verify the realistic possibility of the proposed system architecture, we developed the service prototype.

• Journal of Korea Multimedia Society
• /
• v.25 no.2
• /
• pp.206-214
• /
• 2022

The way to measure the extent of damage from floods and droughts is to identify changes in the extent of water systems. In order to effectively grasp this at a glance, satellite images are used. KOMPSAT-5 uses Synthetic Aperture Radar (SAR) to capture images regardless of weather conditions such as clouds and rain. In this paper, various deep learning models are applied to perform semantic segmentation of the water system in this SAR image and the performance is compared. The models used are U-net, V-Net, U2-Net, UNet 3+, PSPNet, Deeplab-V3, Deeplab-V3+ and PAN. In addition, performance comparison was performed when the data was augmented by applying elastic deformation to the existing SAR image dataset. As a result, without data augmentation, U-Net was the best with IoU of 97.25% and pixel accuracy of 98.53%. In case of data augmentation, Deeplab-V3 showed IoU of 95.15% and V-Net showed the best pixel accuracy of 96.86%.

• Air Cleaning Technology
• /
• v.30 no.1
• /
• pp.1-15
• /
• 2017

• The Magazine of the IEIE
• /
• v.43 no.1
• /
• pp.18-23
• /
• 2016

• The Magazine of the IEIE
• /
• v.43 no.1
• /
• pp.51-56
• /
• 2016

• Korean Journal of Remote Sensing
• /
• v.40 no.1
• /
• pp.71-80
• /
• 2024

In this study, a proposed ensemble learning technique aims to enhance the semantic segmentation performance of images captured by Unmanned Aerial Vehicles (UAVs). With the increasing use of UAVs in fields such as urban planning, there has been active development of techniques utilizing deep learning segmentation methods for land cover segmentation. The study suggests a method that utilizes prominent segmentation models, namely U-Net, DeepLabV3, and Fully Convolutional Network (FCN), to improve segmentation prediction performance. The proposed approach integrates training loss, validation accuracy, and class score of the three segmentation models to enhance overall prediction performance. The method was applied and evaluated on a land cover segmentation problem involving seven classes: buildings,roads, parking lots, fields, trees, empty spaces, and areas with unspecified labels, using images captured by UAVs. The performance of the ensemble model was evaluated by mean Intersection over Union (mIoU), and the results of comparing the proposed ensemble model with the three existing segmentation methods showed that mIoU performance was improved. Consequently, the study confirms that the proposed technique can enhance the performance of semantic segmentation models.