• IEMEK Journal of Embedded Systems and Applications
• /
• v.9 no.5
• /
• pp.261-268
• /
• 2014

Face detection is essential to the full automation of face image processing application system such as face recognition, facial expression recognition, age estimation and gender identification. It is found that local image features which includes Haar-like, LBP, and MCT and the Adaboost algorithm for classifier combination are very effective for real time face detection. In this paper, we present a face detection method using local pixel direction code(PDC) feature and lookup table classifiers. The proposed PDC feature is much more effective to dectect the faces than the existing local binary structural features such as MCT and LBP. We found that our method's classification rate as well as detection rate under equal false positive rate are higher than conventional one.

• ETRI Journal
• /
• v.38 no.2
• /
• pp.347-355
• /
• 2016

In face examinations, gender classification (GC) is one of several fundamental tasks. Recent literature on GC primarily utilizes datasets containing high-resolution images of faces captured in uncontrolled real-world settings. In contrast, there have been few efforts that focus on utilizing low-resolution images of faces in GC. We propose a GC method based on a pixel classifier boosting with modified census transform features. Experiments are conducted using large datasets, such as Labeled Faces in the Wild and The Images of Groups, and standard protocols of GC communities. Experimental results show that, despite using low-resolution facial images that have a 15-pixel inter-ocular distance, the proposed method records a higher classification rate compared to current state-of-the-art GC algorithms.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.223-228
• /
• 2009

We propose an on-line machine learning approach for object recognition, where new images are continuously added and the recognition decision is made without delay. Random forest (RF) classifier has been extensively used as a generative model for classification and regression applications. We extend this technique for the task of building incremental component-based detector. First we employ object descriptor model based on bag of covariance matrices, to represent an object region then run our on-line RF learner to select object descriptors and to learn an object classifier. Experiments of the object recognition are provided to verify the effectiveness of the proposed approach. Results demonstrate that the propose model yields in object recognition performance comparable to the benchmark standard RF, AdaBoost, and SVM classifiers.

• Journal of the Semiconductor & Display Technology
• /
• v.14 no.2
• /
• pp.29-34
• /
• 2015

Traffic lights are common in cities and are important cues for the path planning of intelligent vehicles. In this paper, we propose a robust and efficient algorithm for recognizing traffic lights from video sequences captured by a low cost off-the-shelf camera. Instead of using color information for recognizing traffic lights, a shape based approach is adopted. In learning and detection phase, Histogram of Oriented Gradients (HOG) feature is used and a cascade classifier based on Adaboost algorithm is adopted as the main classifier for locating traffic lights. To decide the color of the traffic light, a technique based on histogram analysis in HSV color space is utilized. Experimental results on several video sequences from typical urban environment prove the effectiveness of the proposed algorithm.

• ETRI Journal
• /
• v.45 no.3
• /
• pp.448-461
• /
• 2023

This article performs a detailed data scrutiny on a chronic kidney disease (CKD) dataset to select efficient instances and relevant features. Data relevancy is investigated using feature extraction, hybrid outlier detection, and handling of missing values. Data instances that do not influence the target are removed using data envelopment analysis to enable reduction of rows. Column reduction is achieved by ranking the attributes through feature selection methodologies, namely, extra-trees classifier, recursive feature elimination, chi-squared test, analysis of variance, and mutual information. These methodologies are ranked via Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) using weight optimization to identify the optimal features for model building from the CKD dataset to facilitate better prediction while diagnosing the severity of the disease. An efficient hybrid ensemble and novel similarity-based classifiers are built using the pruned dataset, and the results are thereafter compared with random forest, AdaBoost, naive Bayes, k-nearest neighbors, and support vector machines. The hybrid ensemble classifier yields a better prediction accuracy of 98.31% for the features selected by extra tree classifier (ETC), which is ranked as the best by TOPSIS.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.2A
• /
• pp.95-104
• /
• 2012

Pattern recognition using ensemble classifiers is composed of strong classifier which consists of many weak classifiers. In this paper, we used feature extraction to organize strong classifier using static camera sequence. The strong classifier is made of weak classifiers which considers environmental factors. So the strong classifier overcomes environmental effect. Proposed method uses binary foreground image by frame difference method and the boosting is used to train crowdedness model and recognize crowdedness using features. Combination of weak classifiers makes strong ensemble classifier. The classifier could make use of potential features from the environment such as shadow and reflection. We tested the proposed system with road sequence and subway platform sequence which are included in "AVSS 2007" sequence. The result shows good accuracy and efficiency on complex environment.

• Journal of Digital Contents Society
• /
• v.18 no.3
• /
• pp.559-565
• /
• 2017

A lot of studies have been conducted to recognize the motion state or behavior of the user using the acceleration sensor built in the smartphone. In this paper, we applied the neural networks to the 3-axis acceleration information of smartphone to study human behavior. There are performance issues in applying time series data to neural networks. We proposed a multi-temporal neural networks which have trained three neural networks with different time windows for feature extraction and uses the output of these neural networks as input to the new neural network. The proposed method showed better performance than other methods like SVM, AdaBoot and IBk classifier for real acceleration data.

• Journal of Information Processing Systems
• /
• v.11 no.2
• /
• pp.165-172
• /
• 2015

The inactive student rate is becoming a major problem in most open universities worldwide. In Indonesia, roughly 36% of students were found to be inactive, in 2005. Data mining had been successfully employed to solve problems in many domains, such as for educational purposes. We are proposing a method for preventing inactive students by mining knowledge from student record systems with several state of the art ensemble methods, such as Bagging, AdaBoost, Random Subspace, Random Forest, and Rotation Forest. The most influential attributes, as well as demographic attributes (marital status and employment), were successfully obtained which were affecting student of being inactive. The complexity and accuracy of classification techniques were also compared and the experimental results show that Rotation Forest, with decision tree as the base-classifier, denotes the best performance compared to other classifiers.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.11a
• /
• pp.497-500
• /
• 2003

본 논문은 계층적 분류기를 제안하여 실시간으로 얼굴 영역을 검출하고, PT(pan-tilt) 카메라를 통해 동적으로 얼굴을 추적할 수 있는 강인한 추적 알고리즘을 구현하고자 한다. 제안된 알고리즘은 분류기 학습, 실시간 얼굴 영역 검출, 추적의 세 단계로 구성된다. 분류기 학습은 AdaBoost 알고리즘을 이용하여, 독특한 얼굴 특징을 추출하는 계층적 분류기를 생성한다. 계층적 분류기는 높은 정확도를 가진 분류기들이 단계적으로 결합됨으로써 우수한 검출 성능으로 수행된다. 실시간 얼굴 영역 검출은 생성된 계층적 분류기를 통해, 빠르고 효율적으로 얼굴 영역을 찾아낸다. 추적은 PT 카메라를 통해 동적으로 검출 영역을 확장시키며, 이전 단계에서 추출된 얼굴 영역의 위치 정보를 이용하여 수행한다. 제안된 알고리즘은 계산의 효율성과 검출 성능을 동시에 증가시키며, 얼굴 검출 수행은 2초당 약 15프레임을 실시간으로 처리한다.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.2E
• /
• pp.74-78
• /
• 2006

This paper proposes an effective method for classifying emotions of the music from its acoustical signals. Two feature sets, timbre and tempo, are directly extracted from the modified discrete cosine transform coefficients (MDCT), which are the output of partial MP3 (MPEG 1 Layer 3) decoder. Our tempo feature extraction method is based on the long-term modulation spectrum analysis. In order to effectively combine these two feature sets with different time resolution in an integrated system, a classifier with two layers based on AdaBoost algorithm is used. In the first layer the MDCT-driven timbre features are employed. By adding the MDCT-driven tempo feature in the second layer, the classification precision is improved dramatically.