• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 추계종합학술대회 논문집(4)
• /
• pp.97-100
• /
• 2000

This paper presents a new approach to the vertex based shape coding technique. The conventional approaches encode objects using a spline method with the same distortion coefficients. The proposed approach, however, classifies the objects based on the object's features, and then applies different distortion values depending on the classified object types. Using this pre-classifying technique, this paper reduces the bit rate and the computational complexity necessary for the encoding process. The performance of the proposed method has been proved by experiments on the various sample Images.

• 한국군사과학기술학회지
• /
• 제25권4호
• /
• pp.381-389
• /
• 2022

Detecting partially occluded objects is difficult due to the appearances and shapes of occluders are highly variable. These variabilities lead to challenges of localizing accurate bounding box or classifying objects with visible object parts. To address these problems, we propose a two-stage part-based attention approach for robust object detection under partial occlusion. First, our part attention network(PAN) captures the important object parts and then it is used to generate weighted object features. Based on the weighted features, the re-weighted object features are produced by our reinforced PAN(RPAN). Experiments are performed on our collected military vehicle dataset and synthetic occlusion dataset. Our method outperforms the baselines and demonstrates the robustness of detecting objects under partial occlusion.

• 융합정보논문지
• /
• 제12권1호
• /
• pp.15-21
• /
• 2022

동영상의 내용 기반 검색에 있어 객체의 움직임에 대한 정보는 내용의 분류와 구분에 있어 중요하게 이용될 수 있다. 특히 사람의 동작을 분석하고 구분하는 것은 검색뿐 아니라 다양한 분야에 활용할 수 있다. 본 논문에서는 객체의 움직임에 따라 변화하는 모양 정보를 이용하여 동작을 표현하고 구분하기 위해 제안된 모양 변화 기술자와 모양 시퀀스의 성능을 높이는 방법을 제안한다. 변화하는 객체의 모양 정보를 더 효율적으로 표현하기 위한 모양 기술자의 선택과 유사도 측정을 위해 사용되는 거리함수의 비교를 통하여 동작 정보의 표현 및 검색 효율을 높일 수 있도록 하였다. 실험을 통하여 제안된 방법이 기존의 방법에 비해 더 효율적으로 동작 정보를 표현하여 검색의 성능을 높일 수 있음을 보였다.

• 자동차안전학회지
• /
• 제12권4호
• /
• pp.23-30
• /
• 2020

Recently, self-driving research has been actively studied in various institutions. Accurate recognition is important because information about surrounding objects is needed for safe autonomous driving. This study mainly deals with the signal processing of LiDAR among sensors for object recognition. LiDAR is a sensor that is widely used for high recognition accuracy. First, we clustered and tracked objects by predicting relative position and speed of objects. The characteristic points of all objects were extracted using point cloud data of each objects through proposed algorithm. The Classification between vehicle and pedestrians is estimated using number of characteristic points and distances among characteristic points. The algorithm for classifying cars and pedestrians was implemented and verified using test vehicle equipped with LiDAR sensors. The accuracy of proposed object classification algorithm was about 97%. The classification accuracy was improved by about 13.5% compared with deep learning based algorithm.

• 정보와 통신
• /
• 제7권3호
• /
• pp.27-34
• /
• 1990

The $(L,\psi)$ feature description on the binary boundary air craft image is introduced of classifying 3-D object (aircraft) identification. Three types for associative matrix memories are employed and tested for their classification performance. The fast association involved in these memories can be implemented using a parallel optical matrix-vector operation. Two associative memories are based on pseudoinverse solutions and the third one is interoduced as a paralell version of a nearest-neighbor classifier. Detailed simulation results for each associative processor are provided.

• 한국멀티미디어학회논문지
• /
• 제25권10호
• /
• pp.1524-1534
• /
• 2022

This study proposes the classification model of Minhwa genre based on object detection of deep learning. To detect unique Korean traditional objects in Minhwa, we construct custom datasets by labeling images using object keywords in Minhwa DB. We train YOLOv5 models with custom datasets, and classify images using predicted object labels result, the output of model training. The algorithm consists of two classification steps: 1) according to the painting technique and 2) genre of Minhwa. Through classifying paintings using this algorithm on the Internet, it is expected that the correct information of Minhwa can be built and provided to users forward.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.877-885
• /
• 2022

Vision-based activity recognition has been widely attempted at construction sites to estimate productivity and enhance workers' health and safety. Previous studies have focused on extracting an individual worker's postural information from sequential image frames for activity recognition. However, various trades of workers perform different tasks with similar postural patterns, which degrades the performance of activity recognition based on postural information. To this end, this research exploited a concept of human-object interaction, the interaction between a worker and their surrounding objects, considering the fact that trade workers interact with a specific object (e.g., working tools or construction materials) relevant to their trades. This research developed an approach to understand the context from sequential image frames based on four features: posture, object, spatial features, and temporal feature. Both posture and object features were used to analyze the interaction between the worker and the target object, and the other two features were used to detect movements from the entire region of image frames in both temporal and spatial domains. The developed approach used convolutional neural networks (CNN) for feature extractors and activity classifiers and long short-term memory (LSTM) was also used as an activity classifier. The developed approach provided an average accuracy of 85.96% for classifying 12 target construction tasks performed by two trades of workers, which was higher than two benchmark models. This experimental result indicated that integrating a concept of the human-object interaction offers great benefits in activity recognition when various trade workers coexist in a scene.

• 정보처리학회논문지:컴퓨터 및 통신 시스템
• /
• 제6권2호
• /
• pp.51-58
• /
• 2017

안전과 보안상의 이유로 감시 카메라의 시장이 확대되고 있으며 이에 대해 영상 인식 및 추적에 관한 연구도 활발히 진행 중에 있으나 인식 및 추적되는 객체의 정보를 획득하여 객체를 식별하는 데는 한계가 있다. 특히, 감시카메라가 활용되는 쇼핑몰, 공항 등과 같은 개방된 공간에서는 다수의 객체들을 식별하기란 더욱 어렵다. 따라서 본 논문에서는 기존의 영상기반 객체 인식 및 추적 시스템에 RFID 기술을 더하여 객체 식별기능을 추가하고자 하였으며 영상 기반과 RFID의 문제 해결을 위해 상호 보완하고자 하였다. 그리하여 시스템의 모듈별 상호작용을 통해 영상기반 객체 인식 및 추적에 실패할 수 있는 문제와 RFID의 인식 오류로 발생할 수 있는 문제에 대한 해결 방안을 제시하였다. 객체의 식별 정도를 4단계로 분류하여 가장 최상의 단계로 객체가 식별이 되도록 시스템을 설계해 식별된 객체의 데이터 신뢰성을 유지할 수 있도록 하였다. 시스템의 효율성 판단을 위해 시뮬레이션 프로그램을 구현하여 이를 입증하였다.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.44-46
• /
• 2022

인공지능을 사용하는 다양한 객체 인식은 기본적으로 평면적인 결과를 보여준다. 물체를 분류하거나 이미지상에 있는 객체가 무엇인지를 확인하는 것을 기초로 한다. 하지만, 원래의 물체는 평면이 아닌 입체적 형태를 가지고 있으며 이미지에서 단순 결과만을 얻기 위한 인식은 상관없지만, 다양한 분야에 활용한다면 부족한 정보가 많다. 본 논문에서는 GAN 알고리즘을 기반으로 한 이미지 생성과 관련하여 중간 결과를 생성하는 Layer의 특성을 활용하여 물체의 다방면의 데이터 생성 방법과 그것이 유의미한지를 확인한다. 기존의 다방면 데이터를 생성하기 위한 하드웨어 및 수집과정에서의 문제점을 일부 해결하고, 몇몇 제한적인 객체에서의 데이터 생성 후 활용이 가능함을 확인한다.

• Journal of information and communication convergence engineering
• /
• 제18권4호
• /
• pp.267-277
• /
• 2020

It is important to understand the exact habitat distribution of endangered species because of their decreasing numbers. In this study, we build a system with a deep learning module that collects the image data of endangered animals, processes the data, and saves the data automatically. The system provides a more efficient way than human effort for classifying images and addresses two problems faced in previous studies. First, specious answers were suggested in those studies because the probability distributions of answer candidates were calculated even if the actual answer did not exist within the group. Second, when there were more than two entities in an image, only a single entity was focused on. We applied an object detection algorithm (YOLO) to resolve these problems. Our system has an average precision of 86.79%, a mean recall rate of 93.23%, and a processing speed of 13 frames per second.