Earticle

도로에서 발생하는 비반복적이며 예측불가능한 일련의 사건을 돌발상황(incident)이라고 하며 이러한 돌발상황이 발생하게 되면 교통류의 정상 흐름이 와해되고 이로써 도로의 용량감소를 일으키며 교통혼잡과 대기오염 등 막대한 사회경제적 손실을 초래한다. 돌발상황으로 인한 피해를 최소화하고자 국내외 각종 교통관리센터에서는 자동 돌발감지 알고리즘에 의한 자동감지 방법을 사용하고 있다. 그러나 현재 운영중인 돌발상황 감지 알고리즘들은 어느 정도의 감지율은 확보하고 있으나 오경보율이 높아 대체적인 성능은 낮은 것으로 판단된다. 유출입램프 수요과다로 인해 도로용량이 다른 구간에 비해 현저히 떨어지는 병목(bottleneck)구간의 경우, 돌발상황이 빈번하게 발생함에도 불구하고 진출입차량으로 인한 대기행렬과 차로변경등의 유사 돌발상황이 발생하여 자동 돌발상황 감지가 더욱 어려운 실정이다. 본 연구에서는 진출입영향권내에서 발생하는 돌발상황을 정확히 감지하기 위해 돌발상황시 혼잡상황 구분을 통한 자동감지 알고리즘을 바탕으로 램프구간의 혼잡 감지시 인접한 본선의 차로를 돌발상황 판단모듈에서 제외함으로써 모형의 성능을 향상시킬 수 있음을 살펴보았다.

A series of accidents, which are non-recurrent and non-anticipated, are called incidents. These incidents make standard traffic flows interrupt, which result in the decrease of road capacity and a number of social and economic costs, such as the traffic congestion and air pollution. In order to prevent the hazard of incidents, domestic and foreign traffic management center are likely to opt auto-sense system with algorithms of auto-incident sense. However, it is evaluated that the algorithms have a low function with frequent wrong alarms, even if they accurately ry to speculate the incidents. In the case of bottleneck which has lack of road capacity, compared with other roads, due to inefficient road structured over-capacity of the demand of on-off ramp, the incidents regularly take place. Nonetheless, it can be more difficult to speculate the auto-incidents sense owing to similar incidents, such as the queue of in-out flows of cars and the change of road line. Throughout this research, the function of the model has improved excluding near road line in the module of the incidents which is based on the auto-incidents algorithms during the sense of the congestion of ramp areas.

차량과 무선 액세스망과의 연동을 통한 새롭고 다양한 멀티미디어 서비스를 제공할 수 있는 차세대 텔레매틱스 기술이 새로운 유망산업으로 소개되고 있다. 초기의 텔레매틱스 기술은 단순히 차량내에서 전자제어장치간의 제어, 모니터링, 내비게이션 등이 주였지만, 최근에는 외부의 무선 액세스망과의 연결을 통하여 모바일 인터넷 서비스, 위치정보 서비스, 긴급구조 서비스, 멀티미디어 서비스 등을 제공하고 있다. 본 논문에서는 이러한 서비스의 활성화와 효율성을 제공하기 위하여 차량내 텔레매틱스 모바일 플랫폼 구조를 제안한다.

Next generation telematics technology, offering new and diverse multimedia service by connecting car and wireless access network, has been presented with promising industry. Telematics technology was simply to monitor and control the in-vehicle devices and to navigate the road in the early days but nowadays, telematics technology has provided the mobile internet access service, LBS service, agent rescue service and multimedia service to us using connection of wireless access network. In this paper, telematics mobile platform architecture is proposed for service promotion and efficiency.

With the rapid development of the Internet and the rapid popularity of the network. Increasing network traffic also leads to frequent abnormal attacks, which seriously affects user information security. Then intrusion detection is studied as an active defense technology. In order to increase the ability of network abnormal traffic detection, a network abnormal traffic detection algorithm based on variational autoencoder is proposed in this paper. In addition to normal data, the algorithm subdivides the intrusion attack types into 4 major categories, which contain 38 specific attacks, each specific attack category as an intrusion label, which will be used as the input of the decoder together with the hidden vector to obtain the classification results more accurately and efficiently and to recover the data features in the incomplete data set. Finally, the trained Deep Belief Nets (DBN) is used for further feature extraction and classification of the data. It is verified that the proposed variational autoencoder-based supervised classification scheme is simple and effective, which is able to identify new attacks well and outperforms other classification schemes in terms of data classification.

To address the high complexity, poor real-time performance, and the prevalence of false positives and false negatives in current algorithms for detecting small-target pollutants on UAV-based building facades, this study proposes SDS-YOLOv8. The spatial pyramid pooling structure in the backbone is enhanced to improve feature representation. DySample is incorporated into the neck to adaptively adjust sampling points based on the image feature distribution. Additionally, the SCAM module is introduced to improve the memory of important information, and the loss function is further optimized. Experimental results demonstrate that the accuracy of the proposed algorithm is significantly improved, exhibiting strong generalization capability. ⓒ2025 The Korean Institute of Communications and Information Sciences. Publishing Services by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

In this letter a new detection scheme combining the conventional Sphere decoders and best detection algorithm is proposed. This algorithm leverages the Sphere decoder (SD) results to use smaller values of in best algorithm to achieve better performance. Post best detection algorithm Darlington pairs of SD- SD, SD- SD, and -1 SD are used to get the final detection results. The proposed best detection algorithm finds out smallest paths by counting and sorting the bits of every candidate, which is much simpler as used in conventional sphere decoders. The computational complexity has been calculated in terms of time taken to visit the average number of K paths taken by Conventional SDs -SDs and Darlington pair of SDs. Extensive Monte Carlo simulations are used to demonstrate that the proposed approach exhibits significant performance gain over conventional SD schemes in terms of bit error rate and computational time. Besides, a complete analytical approach is provided to validate the simulation results.

In recent years, massive MIMO systems have been widely researched to realize high-speed data transmission. Since massive MIMO systems use a large number of antennas, these systems require huge complexity to detect the signal. In this paper, we propose a novel detection method for massive MIMO using parallel detection with maximum likelihood detection with QR decomposition and M-algorithm (QRM-MLD) to reduce the complexity and latency. The proposed scheme obtains an R matrix after permutation of an H matrix and QR decomposition. The R matrix is also eliminated using a Gauss?Jordan elimination method. By using a modified R matrix, the proposed method can detect the transmitted signal using parallel detection. From the simulation results, the proposed scheme can achieve a reduced complexity and latency with a little degradation of the bit error rate (BER) performance compared with the conventional method.

Detection of nodes disseminating false data is a prerequisite for effective deployment of Internet of Vehicles (IoV) services. This work proposed a novel hyper-tuned ensemble Random Forest (Ens. RF) algorithm to detect false basic safety messages in IoV. Performance evaluation was done using the Vehicular Reference Misbehavior (VeReMi) dataset comprising data-centric misbehavior evaluation for vehicular networks. For validation, a comparative analysis of the performance of the proposed “Ens. RF” model, five machine learning algorithms implemented in this work, and state-of-the-art ML models from related literature was presented. The performance metrics considered are time efficiency and validation accuracy for overall misbehavior classification. Also, the results confirmed the irrelevance of data balancing in real-life scenarios. Finally, we assess the performance of our proposed system for detecting each falsification scenario using precision and recall. The result shows that the proposed algorithm outperformed others with a validation accuracy of 99.60% and a negligible 604 misclassifications out of 153,730 points.

This paper proposes a new intrusion detection system (IDS) based on a combination of a multilayer perceptron (MLP) network, and artificial bee colony (ABC) and fuzzy clustering algorithms. Normal and abnormal network traffic packets are identified by the MLP, while the MLP training is done by the ABC algorithm through optimizing the values of linkage weights and biases. The CloudSim simulator and NSL-KDD dataset are used to verify the proposed method. Mean absolute error (MAE), root mean square error (RMSE), and the kappa statistic are considered as evaluation criteria. The obtained results have indicated the superiority of the proposed method in comparison with state-of-the-art methods.

In this paper, we propose an efficient decision rule in order to get better chance to detect the unused spectrum assigned to a licensed user and improve reliability of spectrum sensing performance. Each secondary user receives the signals from the licensed user. And the resulting signals input to an energy detector. Then, each sensing result is combined and used to make a decision whether the primary user is present at the licensed spectrum band or not. In order to make the reliable decision, we apply an efficient decision rule that is called as a majority rule in this paper. The simulation results show that spectrum sensing performance with the proposed decision rule is more reasonable and efficient than that with conventional decision rules.

자율주행 시스템에서 신호등을 포함한 교통 표지판의 정확한 탐지는 안전한 주행과 교통 규정을 준수하기 위해 필수 적이다. 교통 표지판은 속도 제한, 정지 신호, 차선 변경, 보행자 횡단 및 경고 표지 등 중요한 정보를 제공하며, 자율주행 차량은 이를 실시간으로 해석하여 올바른 주행 결정을 내려야 한다. 교통 표지판을 인식하지 못하거나 잘못 해석하는 경우 심각한 안전 위험을 초래할 수 있으므로, 강력한 탐지 알고리즘은 자율주행 기술의 핵심 요소로 간주된다. 본 연구는 YOLOv8 알고리즘을 활용하여 미국 교통 표지판의 인식 및 분류 성능을 향상시키는 것을 목표로 한다. 이 데이터셋은 다양 한 조명 조건, 날씨 변화, 가려짐(occlusion), 그리고 다양한 표지판 각도 등을 포함하여, 모델의 성능을 철저히 평가할 수 있도록 구성되어 있다. YOLOv8 알고리즘과 고품질 Kaggle 데이터셋을 결합함으로써, 본 연구는 교통 표지판 탐지의 정밀 도와 효율성을 향상시키는 것을 목표로 한다. 연구 결과는 자율주행 차량의 환경 인식 기술을 발전시키는 데 기여할 것이며, 복잡한 도로 환경에서도 더욱 안전하고 효과적으로 주행할 수 있도록 돕는다. 또한, 본 연구는 교통 표지판 인식 성능을 높이는 것뿐만 아니라, 자율주행 애플리케이션에서 실시간으로 활용 가능한 최적의 딥러닝 모델 개발에 대한 통찰을 제공할 것이다.

Accurate detection of traffic signs, including traffic lights, is essential for autonomous driving systems to ensure safe navigation and adherence to traffic regulations. Traffic signs provide crucial information, such as speed limits, stop signals, lane changes, pedestrian crossings, and warning signs, all of which self-driving vehicles must interpret in real-time to make informed driving decisions. Failure to recognize or misinterpret traffic signs can lead to critical safety risks, making robust detection algorithms a fundamental component of autonomous vehicle technology. This research explores the application of the YOLOv8 algorithm, a state-of-the-art deep learning-based object detection model, to improve the recognition and classification of US traffic signs. By integrating the YOLOv8 algorithm with a well-curated Kaggle dataset, this study aims to enhance the precision and efficiency of traffic sign detection. The findings of this research are expected to contribute to the advancement of autonomous vehicle perception, improving their ability to navigate complex road environments safely and effectively. This study not only enhances traffic sign recognition but also provides insights into optimizing deep learning models for real-time deployment in self-driving applications.

Background/Objectives Deep learning which main technology in AI has high growth with being applied to field of speech recognition and Image classification. Especially, Deep learning technology in the field of Image classification is being applied as a core technology to Self-driving and crime prevention monitoring system that is recently emerging as the future industry. Methods/Statistical analysis: Various algorithm which is improved and developed CNN being able to do image process is suggested as Deep learning model in image recognition field. In this paper, we introduce various object detection algorithm including CNN. And explore most representative algorithms just R-CNN, Fast R-CNN, Faster R-CNN and difference between versions of YOLO devised to detect and track in real time. Findings: This paper evaluates deep learning algorithm’s performance by comparative analysis about mAP (mean average precision) and FPS (frames per second). In result of performance evaluation, YOLO algorithm is confirmed as that It shows excellent result in speed that detects and recognizes object and accuracy in real time system environment. Finally, we search cases in field of autonomous driving and access control system and home anti-crime system. Improvements/Applications: In this research, we can understand object detection algorithm among speech recognition technologies and proper field in each algorithm, apply security service based on image, recommend proper algorithm in various environment just like autonomous driving and security work, etc.

Soil surface displacement occurs often following timber harvesting. This displacement is caused by canopy removal from the forest area which makes bare soil to be exposed directly to the rainfall. Thus, the displacement may differ in due to precipitation intensity, and this phenomenon results in the displacement prediction to be uncertain. In this study, we investigated on 1) calculating displacement on timber harvested area in research forest by photogrammetry method, and 2) assessing erosion risk of the area by using three different supervised machine learning classification models. Two time-series of point cloud data (PCD) were acquired by photogrammetry method from June 10, 2022 and July 9, 2022, respectively. These two PCD sets were registered by ground control point (GCP). For displacement estimation, DEM of difference (DoD) was calculated by subtracting DEMs from registered PCD sets. A total of nine terrain variables were generated for analyzing the spatial relationship between displacements and topographic features. Erosion risk was analyzed by three different classification algorithms: random forest (RF), extra gradient boost (XGB), logistic regression (LG). Best performed model was RF with overall accuracy (OA) of 72.9% followed by XGB model (72%), and LG model (71.7%). The results of this research indicate that machine learning based classification algorithms have better performances on erosion risk analysis than statistical algorithm through this research.

포트홀은 차량파손과 교통사고 유발 등의 사회문제를 유발시키고 있다. 포트홀을 효율적으로 관리하기 위해서는 빠르게 포트홀을 찾아내는 기술이 가장 중요하다. 기존의 포트홀 탐지 기법은 민원에 의한 수동식 신고방식을 사용하고 있어, 포트홀로 인해 발생하는 문제를 사전에 예방하지 못하고 있다. 최근 포트홀을 저비용으로 빠르게 탐지하기 위하여 영상 카메라를 이용한 연구가 많이 진행되고 있다. 본 논문에서는 사전에 연구되었던 포트홀 탐지 알고리즘의 탐지 정확도를 개선하기 위한 Saliency Map 기반의 알고리즘을 제안한다. 기존 알고리즘은 포트홀이 그림자와 겹쳐있거나 포트홀의 내부 모양이 주변 도로노면과 비슷한 형태를 가지는 등의 복잡한 환경에서 포트홀을 탐지하지 못하는 문제를 가지고 있다. 이러한 문제를 해결하기 위하여 제안하는 알고리즘은 Saliency Map 알고리즘을 이용하여 보다 정확한 포트홀 후보 영역을 찾는다. 제안 알고리즘은 포트홀 후보영역 추출부와 결정부로 구성되며, 실험을 통하여 기존 알고리즘보다 더 높은 탐지 정확도를 가짐을 보인다.

Potholes have caused diverse problems such as wheel damage and car accident. A pothole detection technology is the most important to provide efficient pothole maintenance. The previous pothole detections have been performed by manual reporting methods. Thus, the problems caused by potholes have not been solved previously. Recently, many pothole detection systems based on video cameras have been studied, which can be implemented at low costs. In this paper, we propose a new pothole detection algorithm based on saliency map information in order to improve our previously developed algorithm. Our previous algorithm shows wrong detection with complicated situations such as the potholes overlapping with shades and similar surface textures with normal road surfaces. To address the problems, the proposed algorithm extracts more accurate pothole regions using the saliency map information, which consists of candidate extraction and decision. The experimental results show that the proposed algorithm shows better performance than our previous algorithm.

메타학습 알고리즘인 Adaboost는 단순한 분류기(classifier)인 Haar-like feature 필터와 결 합하여 빠른 속도와 높은 정확도를 나타내기 때문에 많은 객체 검출에 분야에 사용되고 있으며 그 중 특히 얼굴 인식 분야에서 뛰어난 성 능을 나타내고 있다. 그러나 Adaboost 알고리즘은 얼굴인식 시 정면 얼굴과 같은 단일 모델 검출에는 매우 효율적 이지만 측면 얼굴과 정면 얼굴이 혼합된 모델에서의 검출 시 성능이 저하되는 문제점을 가 지고 있다. 이 문제를 해 결하기 위해 본 논문에서는 혼합모델의 검출 성능을 높이기 위해 인스턴스 정보를 저장하여 학습하는 알고리즘인 IBL(Instance Based Learning)를 Haar-like feature 필터와 결합한 알고리즘을 제안한다.

Adaboost is a meta learning algorithm and it is used many area of object detection. In the field of face detection it is used in conjunction with Haar-like feature filtering because of its fast speed and high accuracy. Adaboost shows very effective performance for face detection in single distribution model such as front face but it shows reduced performance on multiple distribution models such as mixture of front face and side face. As a result of reduced performance this paper proposed to improve performance of multiple distribution models using IBL (Instance Based Learning) algorithm combined with the Haar-like feature.

본 연구에서는 대전광역시 주요 간선도로인 유성대로를 대상으로 드론을 통해 취득한 노면 영상데이터를 기반으로 물체탐지알고리즘(Object Detection algorithm) 가운데 Tiny-YOLO-V2와 Faster-RCNN을 활용하여 아스팔트 도로노면의 균열을 인식, 균열유형을 구분하고 실험 결과차 이를 비교하였다. 분석결과, Faster-RCNN의 mAP는 71%이고 Tiny-YOLO-V2의 mAP는 33%로 측 정되었으며, 이는 1stage Detection인 YOLO계열 알고리즘보다 2Stage Detection인 Faster-RCNN 계열의 알고리즘이 도로노면의 균열을 확인하고 분리하는데 더 좋은 성능을 보인다는 것을 확 인하였다. 향후, 드론과 인공지능형 균열검지시스템을 이용한 도로자산관리체계(Infrastructure Asset Management) 구축방안 마련을 통해 효율적이고 경제적인 도로 유지관리 의사결정 지원 시스템 구축 및 운영 환경을 조성할 수 있을 것이라 판단된다.

This paper proposes a new methodology to recognize cracks on asphalt road surfaces using the image data obtained with drones. The target section was Yuseong-daero, the main highway of Daejeon. Furthermore, two object detection algorithms, such as Tiny-YOLO-V2 and Faster-RCNN, were used to recognize cracks on road surfaces, classify the crack types, and compare the experimental results. As a result, mean average precision of Faster-RCNN and Tiny-YOLO-V2 was 71% and 33%, respectively. The Faster-RCNN algorithm, 2Stage Detection, showed better performance in identifying and separating road surface cracks than the Yolo algorithm, 1Stage Detection. In the future, it will be possible to prepare a plan for building an infrastructure asset-management system using drones and AI crack detection systems. An efficient and economical road-maintenance decision-support system will be established and an operating environment will be produced.

이 연구는 유화에서 손상된 부위를 식별하고 추출하는 데 있어 SIFT, ORB, AKAZE, BRIST의 네 가지 특징 감지 알고리즘의 효능을 비교합니다. 다양한 열화 현상을 특징으로 하는 600개의 유화 이미지 데이터 세트를 컴파일하고 전처리했습 니다. 성능 평가 지표에는 감지된 특징의 수, 일치 능력, 일치 비율, 일관성 및 실행 시간이 포함되었습니다. 그 결과 BRIXE는 특징 감지량과 일치 수에서 우수하지만 처리 시간이 증가하는 대가를 치렀습니다. AKAZE와 SIFT는 각각 25.3%와 25%로 가장 높은 일치 비율을 보여 고품질 특징 감지를 시사했습니다. ORB는 이미지 전 반에서 가장 일관된 성능을 보였으며 절대 일치 수에서 2위를 차지했습니다. 이 연 구에 따르면 알고리즘 선택은 특정 작업 요구 사항을 기반으로 해야 합니다: 철저 한 특징 감지를 위한 BRIST, 고품질 일치를 위한 AKAZE 또는 SIFT, 일관되고 효율적인 성능을 위한 ORB. 이 연구는 유화의 손상 평가를 위한 객관적인 방법을 제공하고 잠재적으로 복원 프로세스를 간소화하며 보존 노력의 자원 할당을 최적화 함으로써 예술 보존에 기여합니다.

This study compares the efficacy of four feature detection algorithms: SIFT, ORB, AKAZE, and BRIST in identifying and extracting damaged areas in emulsions. A dataset of 600 emulsified images featuring various deterioration phenomena was compiled and preprocessed. Performance evaluation metrics included the number of detected features, matching ability, matching rate, consistency, and execution time. The results showed that BRIXE was excellent in feature detection and matching number, but at the cost of increasing processing time. AKAZE and SIFT showed the highest matching rates, 25.3% and 25%, respectively, suggesting high-quality feature detection. ORB performed most consistently across images and ranked second in absolute matching number. According to this study, algorithm selection should be based on specific task requirements: BRIST for thorough feature detection, AKAZE or SIFT for high-quality matching, and ORB for consistent and efficient performance. This study contributes to art conservation by providing an objective method for the evaluation of damage in emulsions, potentially simplifying the restoration process, and optimizing the resource allocation of conservation efforts.

도로 노면 파손 탐지는 쾌적한 주행 환경과 안전사고의 예방을 위해 필요하다. 도로 관리 기관은 자동화 기술 기반의 검사 장비와 시스템을 활용하고 있다. 이러한 자동화 기술 중에서 도 도로 노면의 파손을 탐지하는 기술은 중요한 역할을 수행한다. 최근 들어 딥러닝을 이용한 기술에 대한 연구가 활발하게 진행 중이다. 이러한 딥러닝 기술 개발을 위해서는 도로 영상과 라벨 영상이 필요하다. 하지만 라벨 영상을 확보하기 위해서는 많은 시간과 노동력이 요구된 다. 본 논문에서는 이러한 문제를 해결하기 위하여 준지도 학습 기법 중 하나인 적대적 학습 방법을 제안했다. 이를 구현하기 위해서 5,327장의 도로 영상과 1,327장의 라벨 영상을 사용하 여 경량화 심층 신경망 모델을 학습했다. 그리고 이를 400장의 도로 영상으로 실험한 결과 80.54%의 mean intersection over union과 77.85%의 F1 score를 갖는 모델을 개발하였다. 결과적 으로 라벨 영상 없이 도로 영상만을 학습에 추가하여 인식 성능을 향상시킬 수 있는 기술을 개발하였고, 향후 도로 노면 관리를 위한 기술로 활용되길 기대한다.

Road surface damage detection is essential for a comfortable driving environment and the prevention of safety accidents. Road management institutes are using automated technology-based inspection equipment and systems. As one of these automation technologies, a sensor to detect road surface damage plays an important role. For this purpose, several studies on sensors using deep learning have been conducted in recent years. Road images and label images are needed to develop such deep learning algorithms. On the other hand, considerable time and labor will be needed to secure label images. In this paper, the adversarial learning method, one of the semi-supervised learning techniques, was proposed to solve this problem. For its implementation, a lightweight deep neural network model was trained using 5,327 road images and 1,327 label images. After experimenting with 400 road images, a model with a mean intersection over a union of 80.54% and an F1 score of 77.85% was developed. Through this, a technology that can improve recognition performance by adding only road images was developed to learning without label images and is expected to be used as a technology for road surface management in the future.

인구의 감소 및 고령화 사회가 진행되면서 운전자의 평균 연령은 높아지게 된다. 그에 따라 잠재적인 사고의 위험성이 높은 고령 운전자들은 자율 주행형 개인 이동체가 필요하게 된다. 이러한 이동체가 도로 주행 중에 안전성을 확보하기 위하여 여러 장애물에 대응할 기술이 요 구된다. 그 중에서도 주행 중에 마주할 수 있는 차량, 자전거, 사람과 같은 동적 장애물뿐만 아니라 도로 노면의 불량 상태와 같은 정적 장애물을 인식하는 기술이 가장 우선적으로 필요 하다. 이를 위해서 본 논문에서는 두 종류의 장애물을 동시에 탐지할 수 있는 심층 신경망 알 고리즘을 제안했다. 이 알고리즘을 개발하기 위해서 1,418장의 영상을 이용하여 7종의 동적 장 애물에 표기한 annotation data와 도로 노면 파손을 표시한 label 영상을 확보했다. 이를 이용하 여 학습한 결과, 46.22%의 평균 정확도로 동적 장애물을 탐지하고 74.71%의 mean intersection over union으로 도로 노면 파손을 탐지했다. 또한 한 장의 영상을 처리하는데 평균 소요시간은 89ms로 일반 차량보다 느린 개인 이동 차량에 사용하기 적합한 알고리즘을 개발했다. 향후 주 행 중 마주할 있는 도로 장애물을 탐지하는 기술을 활용하여 개인 이동 차량의 주행 안전성이 향상되길 기대한다.

As the population decreases in an aging society, the average age of drivers increases. Accordingly, the elderly at high risk of being in an accident need autonomous-driving vehicles. In order to secure driving safety on the road, several technologies to respond to various obstacles are required in those vehicles. Among them, technology is required to recognize static obstacles, such as poor road conditions, as well as dynamic obstacles, such as vehicles, bicycles, and people, that may be encountered while driving. In this study, we propose a deep neural network algorithm capable of simultaneously detecting these two types of obstacle. For this algorithm, we used 1,418 road images and produced annotation data that marks seven categories of dynamic obstacles and labels images to indicate road damage. As a result of training, dynamic obstacles were detected with an average accuracy of 46.22%, and road surface damage was detected with a mean intersection over union of 74.71%. In addition, the average elapsed time required to process a single image is 89ms, and this algorithm is suitable for personal mobility vehicles that are slower than ordinary vehicles. In the future, it is expected that driving safety with personal mobility vehicles will be improved by utilizing technology that detects road obstacles.