• Journal of Convergence for Information Technology
• /
• v.11 no.10
• /
• pp.45-52
• /
• 2021

In this paper, we proposed an Atrous Residual U-Net (AR-UNet) to improve the segmentation accuracy of semantic segmentation method based on U-Net. The U-Net is mainly used in fields such as medical image analysis, autonomous vehicles, and remote sensing images. The conventional U-Net lacks extracted features due to the small number of convolution layers in the encoder part. The extracted features are essential for classifying object categories, and if they are insufficient, it causes a problem of lowering the segmentation accuracy. Therefore, to improve this problem, we proposed the AR-UNet using residual learning and ASPP in the encoder. Residual learning improves feature extraction ability and is effective in preventing feature loss and vanishing gradient problems caused by continuous convolutions. In addition, ASPP enables additional feature extraction without reducing the resolution of the feature map. Experiments verified the effectiveness of the AR-UNet with Cityscapes dataset. The experimental results showed that the AR-UNet showed improved segmentation results compared to the conventional U-Net. In this way, AR-UNet can contribute to the advancement of many applications where accuracy is important.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.21 no.4
• /
• pp.81-90
• /
• 2018

The 'Hyper-connected society' in which all objects such as people, device, place are connected via networks and share information being realized. As the information and communication environment changes, spatial information also faces a significant challenge. Korean government is striving to meet the social demand for spatial information that will bring 'Hyper-connectivity' such as autonomous vehicles, drones. Until now, however, it has only partially responded to urgent demand and has not prepared a fundamental countermeasure. In order to effectively and actively respond to the demand for spatial information that is needed in the Hyper-connected society, a strategy that can lead to mid- to long-term fundamental changes is needed. The purpose of this study is to analyze the future demand and application characteristics of spatial information confronted with a big paradigm shift called 'Hyper-connected society', and to search spatial information strategy that can cope with the demand of spatial information in future society.

• Journal of the Semiconductor & Display Technology
• /
• v.17 no.4
• /
• pp.21-26
• /
• 2018

Recently, the emergence of 3D cameras, 3D scanners and various cameras including Lidar is expected to be applied to applications such as AR, VR, and autonomous mobile vehicles that deal with 3D data. In Particular, the 3D point cloud data consisting of tens to hundreds of thousands of 3D points is rapidly increased in capacity compared with 2D data, Efficient encoding / decoding technology for smooth service within a limited bandwidth, and efficient service provision technology for differentiating the area of interest and the surrounding area are needed. In this paper, we propose a new quality parameter considering characteristics of 3D point cloud instead of quality change based on assumed video codec in MPEG V-PCC used in 3D point cloud compression, 3D Grid division method and representation for selectively transmitting 3D point clouds according to user's area of interest, and propose a new 3D Manifesto. By using the proposed technique, it is possible to generate more bitrate images, and it is confirmed that the efficiency of network, decoder, and renderer can be increased while selectively transmitting as needed.

• Smart Media Journal
• /
• v.8 no.2
• /
• pp.66-73
• /
• 2019

Recently, object recognition technology using computer vision has attracted attention as a technology to replace sensor-based object recognition technology. It is often difficult to commercialize sensor-based object recognition technology because such approach requires an expensive sensor. On the other hand, object recognition technology using computer vision may replace sensors with inexpensive cameras. Moreover, Real-time recognition is viable due to the growth of CNN, which is actively introduced into other fields such as IoT and autonomous vehicles. Because object recognition model applications demand expert knowledge on deep learning to select and learn the model, such method, however, is challenging for non-experts to use it. Therefore, in this paper, we analyze the structure of deep - learning - based object recognition models, and propose a platform that can automatically select a deep - running object recognition model based on a user 's desired condition. We also present the reason we need to select statistics-based object recognition model through conducted experiments on different models.

• Journal of the Korea Society for Simulation
• /
• v.28 no.1
• /
• pp.93-98
• /
• 2019

In case of unmanned aerial vehicles used in modern society, there has been a problem where a human operator should be still needed to control the UAV because of a lower level of autonomy. In this paper, genetic algorithm is selected as a methodology for the autonomy accomplishment and then we verify a possibility of UAV autonomy by applying the GA. The landing is one of the important classical tasks on aerial vehicle and the lunar Landing is one of the most historical events. Autonomy possibility of computer-simulated UAV is verified by landing autonomy method of a falling body equipped with a propulsion system similar to the lunar Lander. When applying the GA, the genom is encoded only with 4 actions (left-turn, right-turn, thrust, and free-fall) and applied onto the falling body, Then we applied the major operations of GA and achieved a success experiment. A major contribution is to construct a simulated UAV where an autonomy of UAV can be accomplished while minimizing the sensor dependency. Also we implemented a test-bed where the possibility of autonomy accomplishment by applying the GA can be verified.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.169-170
• /
• 2019

Lately, as artificial intelligence becomes a source of global competitiveness, the government is strategically fostering artificial intelligence that is the base technology of future new industries such as autonomous vehicles, drones, and robots. Domestic artificial intelligence research and services have been launched mainly in Naver and Kakao, but their size and level are weak compared to overseas. Recently, deep learning has been conducted in recent years while recording innovative performance in various pattern recognition fields including speech recognition and image recognition. In addition, deep running has attracted great interest from industry since its inception, and global information technology companies such as Google, Microsoft, and Samsung have successfully applied deep learning technology to commercial products and are continuing research and development. Therefore, we will look at artificial intelligence which is attracting attention based on previous research.

• Journal of Auto-vehicle Safety Association
• /
• v.13 no.3
• /
• pp.78-85
• /
• 2021

Recently a traffic accident of heavy duty vehicles under the mandatory installation of ADAS (Advanced Driver Assistance System) is often reported in the media. Heavy duty vehicle accidents are normally occurring a high number of passenger's injury. According to report of Insurance Institute for Highway Safety, FCW (Forward Collision Warning) and AEB (Automatic Emergency Braking) were associated with a statistically significant 12% reduction in the rate of police-reportable crashes per vehicle miles traveled, and a significant 41% reduction in the rear-end crash rate of large trucks. Also many countries around the world, including Korea, are studying the effects of ADAS installation on accident reduction. Traffic accident statistics of passenger vehicle for business purpose in TMACS (Traffic safety information Management Complex System in Korea) tends to remarkably reduce the number of deaths due to the accident (2017(211), 2018(170), 2019(139)), but the number of traffic accidents (2017(8,939), 2018(9,181), 2019(10,095)) increases. In this paper, it is introduced a traffic accident case that could lead to high injury traffic accidents by being equipped with AEB in a bus. AEB reduces accidents and damage in general but malfunction of AEB could occur severe accident. Therefore, proper education is required to use AEB system, simply instead of focusing on developing and installing AEB to prevent traffic accidents. Traffic accident of AEB equipped vehicle may arise a new dispute between a driver's fault and vehicle defect. It is highly recommended to regulate an advanced event data recorder system.

• Journal of Digital Convergence
• /
• v.19 no.6
• /
• pp.251-258
• /
• 2021

In this paper, we proposed an encoder-decoder model utilizing residual learning to improve the accuracy of the U-Net-based semantic segmentation method. U-Net is a deep learning-based semantic segmentation method and is mainly used in applications such as autonomous vehicles and medical image analysis. The conventional U-Net occurs loss in feature compression process due to the shallow structure of the encoder. The loss of features causes a lack of context information necessary for classifying objects and has a problem of reducing segmentation accuracy. To improve this, The proposed method efficiently extracted context information through an encoder using residual learning, which is effective in preventing feature loss and gradient vanishing problems in the conventional U-Net. Furthermore, we reduced down-sampling operations in the encoder to reduce the loss of spatial information included in the feature maps. The proposed method showed an improved segmentation result of about 12% compared to the conventional U-Net in the Cityscapes dataset experiment.

• 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 the Korean Society for Aeronautical & Space Sciences
• /
• v.50 no.1
• /
• pp.47-57
• /
• 2022

UAVs have been deployed various missions such as deception, reconnaissance and attack since they have been applied in battlefield and achieved missions successfully instead of man. In the past, it is impossible for UAVs to conduct autonomous missions or cooperative mission between manned aircraft due to the limitation of the technology. However, theses missions are possible owing to the advance in communication and AI Technology. In this research, we identified the possible cooperative missions between manned and unmanned team based on air-to-air mission. We studied cooperative manned and unmanned tactics about fighter sweep mission which is the core and basic operation among various air-to-air missions. We also developed cooperative tactics of manned and unmanned team by classifying nonstealth and stealth confrontational tactics. Hereafter, we verified the validity of the suggested tactics using computer simulations.