• Title/Summary/Keyword: Precision-recall

Search Result 726, Processing Time 0.031 seconds

Human Activity Classification Using Deep Transfer Learning (딥 전이 학습을 이용한 인간 행동 분류)

  • Nindam, Somsawut;Manmai, Thong-oon;Sung, Thaileang;Wu, Jiahua;Lee, Hyo Jong
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2022.11a
    • /
    • pp.478-480
    • /
    • 2022
  • This paper studies human activity image classification using deep transfer learning techniques focused on the inception convolutional neural networks (InceptionV3) model. For this, we used UFC-101 public datasets containing a group of students' behaviors in mathematics classrooms at a school in Thailand. The video dataset contains Play Sitar, Tai Chi, Walking with Dog, and Student Study (our dataset) classes. The experiment was conducted in three phases. First, it extracts an image frame from the video, and a tag is labeled on the frame. Second, it loads the dataset into the inception V3 with transfer learning for image classification of four classes. Lastly, we evaluate the model's accuracy using precision, recall, F1-Score, and confusion matrix. The outcomes of the classifications for the public and our dataset are 1) Play Sitar (precision = 1.0, recall = 1.0, F1 = 1.0), 2), Tai Chi (precision = 1.0, recall = 1.0, F1 = 1.0), 3) Walking with Dog (precision = 1.0, recall = 1.0, F1 = 1.0), and 4) Student Study (precision = 1.0, recall = 1.0, F1 = 1.0), respectively. The results show that the overall accuracy of the classification rate is 100% which states the model is more powerful for learning UCF-101 and our dataset with higher accuracy.

The Design an Implementation of Content-based Image Retrieval System Using Color Features (칼라 특징을 이용한 내용기반 화상검색시스템의 설계 및 구현)

  • 정원일;박정찬;최기호
    • Journal of the Korean Institute of Telematics and Electronics B
    • /
    • v.33B no.6
    • /
    • pp.111-118
    • /
    • 1996
  • A content-based image retrieval system is designed and implemetned using the color featurees which are histogram intersection and color pairs. The preprocessor for the image retrieval manage linearly the existing HSI(hue, saturation, saturation, intensity). Hue and intensity histogram thresholding for each color attribute is performed to split the chromatic and achromatic regions respectively. Grouping te indexes produced by the histogram intersection is used to save the retrieval times. Each image is divided into the cells of 32$\times$32 pixels, and color pairs are used to represent the query during retrievals. The recall/precision of histogram intersection is 0.621/0.663 and recall/precision of color pairs is 0.438/0.536. And recall/precision of proposed method is 0.765/0.775/. It is shown that the proposed method using histogram intersection and color pairs improves the retrieval rates.

  • PDF

Content-based Image Reterieval Using Color and Chain Code (색상과 Chain Code를 이용한 내용기반 영상검색)

  • 정성호;이상렬;황병곤
    • Proceedings of the Korea Society for Industrial Systems Conference
    • /
    • 1999.12a
    • /
    • pp.193-198
    • /
    • 1999
  • 본 논문에서는 영상의 내용을 나타내는 색상과 체인 코드에 기반한 복잡도와 영역 색정보를 이용한 내용 기반 영상 검색을 결합한 시스템을 구현하였다. 실험 대상으로 선택한 꽃 영상의 경우 대부분의 인식 대상 객체가 중앙에 있을 경우를 고려하여, 영상을 중앙 영역과 전체 영역으로 구분하고, 각각의 영역에서 두 개의 히스토그램을 생성한다. 그리고 전체 영역에 대한 기준치를 구하고 chain code글 이용한 복잡도를 구하였다. 중앙영역과 전체영역의 히스토그램 인터섹션을 이용한 검색을 실험하였고, 영역 색상과 복잡도를 결합한 검색도 또한 실험하였다. 기존의 히스토그램 인터섹션의 경우 Precision/Recall이 0.370/0.60인데 비해 영역 색상 히스토그램을 인터섹션한 경우의 Precision/Recall은 0.69/0.76이고 복잡도를 결합한 경우의 Precision/Recall은 0.92/0.80를 얻음으로써, 제안된 방식의 검색이 비교적 효율적임을 보였다.

  • PDF

Comparison of Retrieval Effectiveness between Precision/Recall and Wilcoxon Test (내용기반 영상 검색 유효성을 측정하는 방법들에 대한 비교 (Precision/Recall vs. Wilcoxon 순위 방법))

  • 장순자;김형중;여인권
    • Proceedings of the Korea Multimedia Society Conference
    • /
    • 2001.11a
    • /
    • pp.453-457
    • /
    • 2001
  • 영상검색은 문자위주의 검색을 한계로, 그 내용에 기반 한 특징벡터를 이용하여 검색을 수행한다. 특징벡터간의 거리를 계산하고 그 값들에 순위를 매긴다. 이렇게 얻어진 순위 값들을 가지고 사용된 검색방법의 유효성을 검사하는데 Recall/Precision 방법이 이용되고 있다. 질의영상과 같은 군에 속하는 영상이 얼마나 검출되었는지를 검출된 영상에 기반하여 계산하거나, 영상 군에 기반하여 계산하는 방법들이다. 그러나, 검출되는 순위값의 범위를 정하고, 그 범위 내에 속하는 만족하는 값의 개수를 세는 방법을 이용한다. 따라서 주어진 두 샘플의 전체적인 경향을 비교하지는 못한다. 본 연구에서는 순위를 이용하여 비교하고자 하는 두 샘플의 순위들을 결합하여 순위를 매기고 각 샘플들에 매겨진 순위값들의 평균과 분산을 이용하여 각 샘플들을 전체적으로 비교할 수 있다.

  • PDF

Content-based Image Retrieval Using Region Color and Keyword (영역 색상과 키워드를 이용한 내용기반 영상검색)

  • 김지영;정성호;황병곤
    • Proceedings of the Korea Society for Industrial Systems Conference
    • /
    • 1999.05a
    • /
    • pp.68-74
    • /
    • 1999
  • 본 논문에서는 영상의 내용을 나타내는 키워드를 이용하는 기존의 텍스트 기반 영상 검색과 영역 색상 정보를 이용한 내용 기반 영상 검색을 결합한 시스템을 구현함으로서, 보다 효과적인 영상 검색을 할 수 있도록 하였다. 영상의 크기는 입력된 원 영상을 사용하였으며, 색상 정보 추출에 있어 HSI 공간으로 변환하여 256개의 칼라로 양자화하였다. 보통의 정지 영상의 경우 대부분의 객체가 중앙에 있을 경우를 고려하여, 영상을 중앙 영역과 배경 영역으로 구분하고, 각각의 영역에서 두 개의 히스토그램을 생성한다. 중앙 영역과 배경영역의 히스토그램 인터섹션을 이용한 검색을 실험하였고, 영역색상과 기존의 키워드를 결합한 검색도 또한 실험하였다. 기존의 히스토그램 인터섹션의 경우 Precision/Recall이 0.34/0.60인데 비해 영역 색상 히스토그램을 인터섹션한 경우의 Precision/Recall은 0.69/0.76이고 키워드를 결합한 경우의 Precision/Recall은 0.92/0.80를 얻음으로써, 제안된 방식의 검색이 비교적 효율적임을 보였다.

  • PDF

A Method of Bank Telemarketing Customer Prediction based on Hybrid Sampling and Stacked Deep Networks (혼성 표본 추출과 적층 딥 네트워크에 기반한 은행 텔레마케팅 고객 예측 방법)

  • Lee, Hyunjin
    • Journal of Korea Society of Digital Industry and Information Management
    • /
    • v.15 no.3
    • /
    • pp.197-206
    • /
    • 2019
  • Telemarketing has been used in finance due to the reduction of offline channels. In order to select telemarketing target customers, various machine learning techniques have emerged to maximize the effect of minimum cost. However, there are problems that the class imbalance, which the number of marketing success customers is smaller than the number of failed customers, and the recall rate is lower than accuracy. In this paper, we propose a method that solve the imbalanced class problem and increase the recall rate to improve the efficiency. The hybrid sampling method is applied to balance the data in the class, and the stacked deep network is applied to improve the recall and precision as well as the accuracy. The proposed method is applied to actual bank telemarketing data. As a result of the comparison experiment, the accuracy, the recall, and the precision is improved higher than that of the conventional methods.

Design of WWW IR System Based on Keyword Clustering Architecture (색인어 말뭉치 처리를 기반으로 한 웹 정보검색 시스템의 설계)

  • 송점동;이정현;최준혁
    • The Journal of Information Technology
    • /
    • v.1 no.1
    • /
    • pp.13-26
    • /
    • 1998
  • In general Information retrieval systems, improper keywords are often extracted and different search results are offered comparing to user's aim bacause the systems use only term frequency informations for selecting keywords and don't consider their meanings. It represents that improving precision is limited without considering semantics of keywords because recall ratio and precision have inverse proportion relation. In this paper, a system which is able to improve precision without decreasing recall ratio is designed and implemented, as client user module is introduced which can send feedbacks to server with user's intention. For this purpose, keywords are selected using relative term frequency and inverse document frequency and co-occurrence words are extracted from original documents. Then, the keywords are clustered by their semantics using calculated mutual informations. In this paper, the system can reject inappropriate documents using segmented semantic informations according to feedbacks from client user module. Consequently precision of the system is improved without decreasing recall ratio.

  • PDF

Developing a Pedestrian Satisfaction Prediction Model Based on Machine Learning Algorithms (기계학습 알고리즘을 이용한 보행만족도 예측모형 개발)

  • Lee, Jae Seung;Lee, Hyunhee
    • Journal of Korea Planning Association
    • /
    • v.54 no.3
    • /
    • pp.106-118
    • /
    • 2019
  • In order to develop pedestrian navigation service that provides optimal pedestrian routes based on pedestrian satisfaction levels, it is required to develop a prediction model that can estimate a pedestrian's satisfaction level given a certain condition. Thus, the aim of the present study is to develop a pedestrian satisfaction prediction model based on three machine learning algorithms: Logistic Regression, Random Forest, and Artificial Neural Network models. The 2009, 2012, 2013, 2014, and 2015 Pedestrian Satisfaction Survey Data in Seoul, Korea are used to train and test the machine learning models. As a result, the Random Forest model shows the best prediction performance among the three (Accuracy: 0.798, Recall: 0.906, Precision: 0.842, F1 Score: 0.873, AUC: 0.795). The performance of Artificial Neural Network is the second (Accuracy: 0.773, Recall: 0.917, Precision: 0.811, F1 Score: 0.868, AUC: 0.738) and Logistic Regression model's performance follows the second (Accuracy: 0.764, Recall: 1.000, Precision: 0.764, F1 Score: 0.868, AUC: 0.575). The precision score of the Random Forest model implies that approximately 84.2% of pedestrians may be satisfied if they walk the areas, suggested by the Random Forest model.

A Study on Unconsciousness Authentication Technique Using Machine Learning in Online Easy Payment Service (온라인 간편 결제 환경에서 기계학습을 이용한 무자각 인증 기술 연구)

  • Ryu, Gwonsang;Seo, Changho;Choi, Daeseon
    • Journal of the Korea Institute of Information Security & Cryptology
    • /
    • v.27 no.6
    • /
    • pp.1419-1429
    • /
    • 2017
  • Recently, environment based authentication technique had proposed reinforced authentication, which generating statistical model per user after user login history classifies into account takeover or legitimate login. But reinforced authentication is likely to be attacked if user was not attacked in past. To improve this problem in this paper, we propose unconsciousness authentication technique that generates 2-Class user model, which trains user's environmental information and others' one using machine learning algorithms. To evaluate performance of proposed technique, we performed evasion attacks: non-knowledge attacker that does not know any information about user, and sophisticated attacker that only knows one information about user. Experimental results against non-knowledge attacker show that precision and recall of Class 0 were measured as 1.0 and 0.998 respectively, and experimental results against sophisticated attacker show that precision and recall of Class 0 were measured as 0.948 and 0.998 respectively.

Adaptive Ontology Matching Methodology for an Application Area (응용환경 적응을 위한 온톨로지 매칭 방법론에 관한 연구)

  • Kim, Woo-Ju;Ahn, Sung-Jun;Kang, Ju-Young;Park, Sang-Un
    • Journal of Intelligence and Information Systems
    • /
    • v.13 no.4
    • /
    • pp.91-104
    • /
    • 2007
  • Ontology matching technique is one of the most important techniques in the Semantic Web as well as in other areas. Ontology matching algorithm takes two ontologies as input, and finds out the matching relations between the two ontologies by using some parameters in the matching process. Ontology matching is very useful in various areas such as the integration of large-scale ontologies, the implementation of intelligent unified search, and the share of domain knowledge for various applications. In general cases, the performance of ontology matching is estimated by measuring the matching results such as precision and recall regardless of the requirements that came from the matching environment. Therefore, most research focuses on controlling parameters for the optimization of precision and recall separately. In this paper, we focused on the harmony of precision and recall rather than independent performance of each. The purpose of this paper is to propose a methodology that determines parameters for the desired ratio of precision and recall that is appropriate for the requirements of the matching environment.

  • PDF