• 대한임베디드공학회논문지
• /
• 제16권5호
• /
• pp.209-214
• /
• 2021

Panoptic segmentation, which is now widely used in computer vision such as medical image analysis, and autonomous driving, helps understanding an image with holistic view. It identifies each pixel by assigning a unique class ID, and an instance ID. Specifically, it can classify 'thing' from 'stuff', and provide pixel-wise results of semantic prediction and object detection. As a result, it can solve both semantic segmentation and instance segmentation tasks through a unified single model, producing two different contexts for two segmentation tasks. Semantic segmentation task focuses on how to obtain multi-scale features from large receptive field, without losing low-level features. On the other hand, instance segmentation task focuses on how to separate 'thing' from 'stuff' and how to produce the representation of detected objects. With the advances of both segmentation techniques, several panoptic segmentation models have been proposed. Many researchers try to solve discrepancy problems between results of two segmentation branches that can be caused on the boundary of the object. In this survey paper, we will introduce the concept of panoptic segmentation, categorize the existing method into two representative methods and explain how it is operated on two methods: top-down method and bottom-up method. Then, we will analyze the performance of various methods with experimental results.

• 한국건축시공학회:학술대회논문집
• /
• 한국건축시공학회 2021년도 봄 학술논문 발표대회
• /
• pp.40-41
• /
• 2021

The construction industry has the highest occupational accidents/injuries among all industries. Korean government installed surveillance camera systems at construction sites to reduce occupational accident rates. Construction safety managers are monitoring potential hazards at the sites through surveillance system; however, the human capability of monitoring surveillance system with their own eyes has critical issues. Therefore, this study proposed to build a deep learning-based safety monitoring system that can obtain information on the recognition, location, identification of workers and heavy equipment in the construction sites by applying multiple-object tracking with instance segmentation. To evaluate the system's performance, we utilized the MS COCO and MOT challenge metrics. These results present that it is optimal for efficiently automating monitoring surveillance system task at construction sites.

• International Journal of Internet, Broadcasting and Communication
• /
• 제15권3호
• /
• pp.66-72
• /
• 2023

Little attention appears to have been paid to the relevance of learning a good representation function in solving long tail tasks. Therefore, we propose a new loss function to ensure a good representation is learnt while learning to classify. We call this loss function Triplet Class-Wise Difficulty-Based (TriCDB-CE) Loss. It is a combination of the Triplet Loss and Class-wise Difficulty-Based Cross-Entropy (CDB-CE) Loss. We prove its effectiveness empirically by performing experiments on three benchmark datasets. We find improvement in accuracy after comparing with some baseline methods. For instance, in the CIFAR-10-LT, 7 percentage points (pp) increase relative to the CDB-CE Loss was recorded. There is more room for improvement on Places-LT.

• 한국농공학회논문집
• /
• 제65권6호
• /
• pp.15-22
• /
• 2023

In this study, We conduct a comparative study of deep learning-based classification of agricultural field attributes using Tagged Image File (TIF) and Enhanced Compression Wavelet (ECW) images. The goal is to interpret and classify the attributes of agricultural fields by analyzing the differences between these two image formats. "FarmMap," initiated by the Ministry of Agriculture, Food and Rural Affairs in 2014, serves as the first digital map of agricultural land in South Korea. It comprises attributes such as paddy, field, orchard, agricultural facility and ginseng cultivation areas. For the purpose of comparing deep learning-based agricultural attribute classification, we consider the location and class information of objects, as well as the attribute information of FarmMap. We utilize the ResNet-50 instance segmentation model, which is suitable for this task, to conduct simulated experiments. The comparison of agricultural attribute classification between the two images is measured in terms of accuracy. The experimental results indicate that the accuracy of TIF images is 90.44%, while that of ECW images is 91.72%. The ECW image model demonstrates approximately 1.28% higher accuracy. However, statistical validation, specifically Wilcoxon rank-sum tests, did not reveal a significant difference in accuracy between the two images.

• 한국컴퓨터정보학회논문지
• /
• 제10권5호
• /
• pp.343-352
• /
• 2005

기존의 초등과학 CAI나 웹기반 학습시스템의 문제점은 교수자의 학습 설계에 따라 획일적인 코스로 구성되어 있고 개인의 학습 특성을 고려하지 않은 채로 설계되었으며 학습도중에 발생되는 학습자의 의견이나 생각을 시스템에 반영 할 수 없어 학습 참여의욕과 학습 동기유발을 떨어뜨려 학습 효율을 높이지 못한다는 점이다. 따라서 학습자 개인의 특성에 맞는 학습 환경을 제공해 주며 학습자의 적극적인 학습참여와 학습 동기유발을 촉진하는 교수-학습 시스템이 필요하다. 본 논문은 캐릭터 교사와 학습자의 대화를 통해 적극적인 학습참여의 기회를 부여하고 지속적인 학습 흥미와 동기유발을 위해 다양한 학습 자료를 제시해 주는 캐릭터 기반의 대화형 교수-학습 시스템을 설계하였다.

• 한국멀티미디어학회논문지
• /
• 제10권11호
• /
• pp.1516-1522
• /
• 2007

FPA(Fixed Partition Averaging) 기법은 기억공간의 효율적인 사용과 분류성능의 향상을 위하여 제안되었던 메모리 기반 추론 기법으로 대상 패턴 공간을 분할 한 후 대표 패턴을 추출하여 분류 기준 패턴으로 사용한다. 이 기법은 메모리 사용 효율과 분류 성능 면에서 우수한 결과를 보인다. 그러나 여러 클래스가 혼합된 분할패턴공간의 경우에 원래의 패턴들을 그대로 저장하여 메모리와 분류성능에 부담으로 작용하는 문제점을 가지고 있다. 본 논문에서는 여러 클래스가 혼합된 분할공간에서 패턴비율을 고려하여 고정분할을 반복적으로 실행하여 초월평면을 생성하고 분류하는 반복적 고정분할평균기법을 제안한다. 본 논문에서 제안한 기법은 기존의 k-NN 기법과 비교하여 현저하게 줄어든 대표패턴을 이용하여 유사한 분류 성능을 보여주며, NGE 이론을 구현한 EACH 시스템과 FPA 기법 등과 비교하여 탁월한 분류 성능을 보여준다.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2022년도 추계학술대회
• /
• pp.171-174
• /
• 2022

Weakly-supervised learning is a widely adopted approach in video anomaly detection whereby only video labels are utilized instead of expensive frame-level annotations. Since the success of multi-instance learning (MIL), almost all recent approaches are based on maximizing the margin between the set of abnormal video snippets and those of normal video snippets. In this work, we present a simple contrastive approach for weakly supervised video anomaly detection (WS-VAD) with aims to enhance the performance of existing models. The method is generic in nature and introduces a loss function to encourage attraction of output features from the same video class and repel those from different video classes. Experimental results demonstrate our method can be applied to existing algorithms to improve detection accuracy in public video anomaly dataset.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권7호
• /
• pp.3638-3653
• /
• 2019

Designing a user interface is important because the user interface determines the level of physical and mental engagement of the user resulting in their level of learning. This paper investigated how physical engagement through a different user interfaces is associated with fun and learning and presented a theoretical physical engagement model called, PEM, developed based on an empirical user study. The PEM model describes how a game user interface is associated with the level of fun and learning, particularly in playing a full body engaged game. There are many different types of games but the Wii Tennis, an embodied interactive game, was chosen as an instance of full body engaged game. A user study with 32 participant's age ranged from 21 to 40 years old revealed that there is a positive correlation between both fun and learning and the level of physical engagement through two different user interfaces. The results of the study showed that the extent of fun and learning are associated with the physical engagement of the player through an interface. As an implication from the study, the result recommend that the level of user engagement is realized by an effective user interface, and the level of physical engagement is determined by the level of authenticity bridged by the user interface.

• International Journal of Computer Science & Network Security
• /
• 제23권7호
• /
• pp.202-209
• /
• 2023

Android malware is now on the rise, because of the rising interest in the Android operating system. Machine learning models may be used to classify unknown Android malware utilizing characteristics gathered from the dynamic and static analysis of an Android applications. Anti-virus software simply searches for the signs of the virus instance in a specific programme to detect it while scanning. Anti-virus software that competes with it keeps these in large databases and examines each file for all existing virus and malware signatures. The proposed model aims to provide a machine learning method that depend on the malware detection method for Android inability to detect malware apps and improve phone users' security and privacy. This system tracks numerous permission-based characteristics and events collected from Android apps and analyses them using a classifier model to determine whether the program is good ware or malware. This method used the machine learning techniques KNN-SVM, DBN, and GRU in which help to find the accuracy which gives the different values like KNN gives 87.20 percents accuracy, SVM gives 91.40 accuracy, Naive Bayes gives 85.10 and DBN-GRU Gives 97.90. Furthermore, in this paper, we simply employ standard machine learning techniques; but, in future work, we will attempt to improve those machine learning algorithms in order to develop a better detection algorithm.

• 방송공학회논문지
• /
• 제27권1호
• /
• pp.44-55
• /
• 2022

농업 관련 방송 콘텐츠에서 과일에 대한 자동적인 정보 제공을 위해서 대상 과일의 인스턴스 영상 분할이 요구된다. 또한, 해당 과일에 대한 3차원 자세에 대한 정보 제공도 의미있게 사용될 수 있다. 본 논문에서는 영상 콘텐츠에서 토마토에 대한 정보를 제공하는 연구를 다룬다. 인스턴스 영상 분할 기법을 학습하기 위해서는 다량의 데이터가 필요하지만 충분한 토마토 학습데이터를 얻기는 힘들다. 따라서 적은 양의 실사 영상을 바탕으로 데이터 증대기법을 통해 학습 데이터를 생성하였다. 실사 영상만을 통한 학습 결과 정확도에 비해서, 전경과 배경을 분리해서 만들어진 합성 영상을 통해 학습한 결과, 기존 대비 성능이 향상되는 것을 확인하였다. 영상 전처리 기법들을 활용해서 만들어진 영상을 사용한 데이터 증대 영상의 학습 결과, 전경과 배경을 분리한 합성 영상보다 높은 성능을 얻는 것을 확인하였다. 객체 검출 후 자세 추정을 하기 위해 RGB-D 카메라를 이용하여 포인트 클라우드를 획득하였고 최소제곱법을 이용한 실린더 피팅을 진행하였고, 실린더의 축 방향을 통해 토마토 자세를 추정하였다. 우리는 다양한 실험을 통해서 대상 객체에 대한 검출, 인스턴스 영상 분할, 실린더 피팅의 결과가 의미있게 나타난다는 것을 보였다.