• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.2
• /
• pp.483-503
• /
• 2014

In this paper, human action recognition using pyramid histograms of oriented gradients and collaborative multi-task learning is proposed. First, we accumulate global activities and construct motion history image (MHI) for both RGB and depth channels respectively to encode the dynamics of one action in different modalities, and then different action descriptors are extracted from depth and RGB MHI to represent global textual and structural characteristics of these actions. Specially, average value in hierarchical block, GIST and pyramid histograms of oriented gradients descriptors are employed to represent human motion. To demonstrate the superiority of the proposed method, we evaluate them by KNN, SVM with linear and RBF kernels, SRC and CRC models on DHA dataset, the well-known dataset for human action recognition. Large scale experimental results show our descriptors are robust, stable and efficient, and outperform the state-of-the-art methods. In addition, we investigate the performance of our descriptors further by combining these descriptors on DHA dataset, and observe that the performances of combined descriptors are much better than just using only sole descriptor. With multimodal features, we also propose a collaborative multi-task learning method for model learning and inference based on transfer learning theory. The main contributions lie in four aspects: 1) the proposed encoding the scheme can filter the stationary part of human body and reduce noise interference; 2) different kind of features and models are assessed, and the neighbor gradients information and pyramid layers are very helpful for representing these actions; 3) The proposed model can fuse the features from different modalities regardless of the sensor types, the ranges of the value, and the dimensions of different features; 4) The latent common knowledge among different modalities can be discovered by transfer learning to boost the performance.

• Journal of KIISE
• /
• v.43 no.5
• /
• pp.531-540
• /
• 2016

Development of various sensors attached to mobile and wearable devices has led to increasing recognition of current context-based service to the user. In this study, we proposed a probabilistic model for recognizing user's food intake context, which can occur in a great variety of contexts. The model uses low-level sensor data from mobile and wrist-wearable devices that can be widely available in daily life. To cope with innate complexity and fuzziness in high-level activities like food intake, a context model represents the relevant contexts systematically based on 4 components of activity theory and 5 W's, and tree-structured Bayesian network recognizes the probabilistic state. To verify the proposed method, we collected 383 minutes of data from 4 people in a week and found that the proposed method outperforms the conventional machine learning methods in accuracy (93.21%). Also, we conducted a scenario-based test and investigated the effect contribution of individual components for recognition.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.4
• /
• pp.40-49
• /
• 1994

In this paper, we propose methods for improving the performance of word recognition system. The ray stratey of the first method is to apply the inertia to the feature vector sequences of speech signal to stabilize the transitions between VQ cdoes. The second method is generating the new observation probabilities using the transition matrix of VQ codes as weights at the observation probability of the output symbol, so as to take into account the time relation between neighboring frames in DHMM. By applying the inertia to the feature vector sequences, we can reduce the overlapping of probability distribution of the response paths for each word and stabilize state transitions in the HMM. By using the transition matrix of VQ codes as weights in conventional DHMM. we can divide the probability distribution of feature vectors more and more, and restrict the feature distribution to a suitable region so that the performance of recognition system can improve. To evaluate the performance of the proposed methods, we carried out experiments for 50 DDD area names. As a result, the proposed methods improved the recognition rate by $4.2\%$ in the speaker-dependent test and $12.45\%$ in the speaker-independent test, respectively, compared with the conventional DHMM.

• Journal of Korea Multimedia Society
• /
• v.11 no.3
• /
• pp.301-314
• /
• 2008

The existing event recognition is accomplished with the limited systematic foundation, and thus much longer learning time is needed for emergency scenario interpretation due to large scale of probability data. In this paper, we propose a method for nile-based event recognition of an independent object(human) which extract a feature vectors from the object and analyze the behavior pattern of each object and interpretation of emergency scenarios using a probability and object's events. The event rule of an independent object is composed of the Primary-event, Move-event, Interaction-event, and 'FALL DOWN' event and is defined through feature vectors of the object and the segmented motion orientated vector (SMOV) in which the dynamic Bayesian network is applied. The emergency scenario is analyzed using current state of an event and its post probability. In this paper, we define diversified events compared to that of pre-existing method and thus make it easy to expand by increasing independence of each events. Accordingly, semantics information, which is impossible to be gained through an.

• The Journal of the Korea Contents Association
• /
• v.18 no.2
• /
• pp.370-377
• /
• 2018

Focus level has been important mental state in user study. Cardiac response has been related to focus and less clarified. The study was to determine cardiac parameters for recognizing focus level. The sixty participants were asked to play shooting game designed to control two focus levels. Electrocardiogram was measured during task. The parameters of time domain and frequency domain were determined from ECG. As a result of independent t-test, RRI, SDNN, rMSSD and pNN50 of time domain indicator were statistically significant in recognizing focus level. LF, HF, lnLF and lnHF of frequency domain were observed to be significant indicator. The rule base for recognition has been developed by the combination of RRI, rMSSD and lnHF. The rule base has been verified from another sixty data samples. The recognition accuracy were 95%. This study proposed significant cardiac indicators for recognizing focus level. The results provides objective measurement of focus in user interaction design in the fields of contents industry and service design.

• Korean Journal of Human Ecology
• /
• v.16 no.6
• /
• pp.1083-1095
• /
• 2007

This study had been done to investigate that early childhood teacher's recognition and demand on children's language education and 20 early childhood teachers were interviewed individually who work at state-owned/ public-owned/ private-owned kindergardens residing G, I, and K cities in Jeollabuk-do. First, the purpose of language education was recognized on the formations of essence, concept, expertise, technique and attitude toward language. Second, the contents of language education must be selected by children's experience that they encounter in ordinary life based on oral language and written language. Besides, early childhood teachers strongly felt the necessity of new contents of language education, although they thought of insufficiency of their knowledge on the issue. Third, the method of language education was mainly accomplished by teaching material and objects. Besides, they were aware of looking for new organized teaching methods and also concerned of the importances of teacher's attitude and group formation method. Fourth, the evaluation of language education must be acquired by desirable evaluation method that was based on the recognition of children's unrealistic language capabilities, even though they had recognized the difficulty to do because of knowledge insufficiency. They also showed the tendency of negligence on the evaluation of language education. Fifth, the required facts for early childhood teachers on language education were development and supply of teaching materials, demand on teacher's education and appropriate evaluation method, and cognitive changes on language education by public toward the written language.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.2
• /
• pp.173-179
• /
• 2004

Due to the insufficiency of training data in large vocabulary continuous speech recognition, similar context dependent phones can be clustered by decision trees to share the data. When the decision trees are built and used to predict unseen triphones, a phonetic question set is required. The phonetic question set, which contains categories of the phones with similar co-articulation effects, is usually generated by phonetic or linguistic experts. This knowledge-based approach for generating phonetic question set, however, may reduce the homogeneity of the clusters. Moreover, the experts must adjust the question sets whenever the language or the PLU (phone-like unit) of a recognition system is changed. Therefore, we propose a data-driven method to automatically generate phonetic question set. Since the proposed method generates the phone categories using speech data distribution, it is not dependent on the language or the PLU, and may enhance the homogeneity of the clusters. In large vocabulary speech recognition experiments, the proposed algorithm has been found to reduce the error rate by 14.3%.

• The Korean Journal of Emergency Medical Services
• /
• v.2 no.1
• /
• pp.58-72
• /
• 1998

This study is implemented in the Department of Emergency Medical Technology of the Cheju Halla College in to evaluate behavioral changes in the recognition of the importance of the first aid, coping ability in the field, and competence of the first aid skills after taking the in-service program. The in-service program was hold from December 14 to December 22, 1998 under the title of "1998 First Aid and Emergency Rescue Training" for 176 drivers. The Questionnaire was distributed to 176 drivers, among them, 88 drivers took in-service program and 88 ones did not. After the data analysis, following conclusions we re made. 1. There are no significant statistical differences among demographic factors such as the level of education, age, career, and marital state of the subjects in the recognition of the importance of the first aid and competence of the first aid skills. 2. The drivers, who were the in-service program, significantly higher score in educated in represented all the area of recognition of the importance of first aid than non-educated drivers. 3. The drivers, who were educated in the in-service program, showed also significantly high score in competence of the first aid skills. According to the conclusions, there were significant change made In the recognition of the importance of the first aid and competence of the first aid skills after the in-service program. Therefore, nationwide emergency training program should be considered to improve emergency care ability of the nation. To make the program more effective for drivers, the in-service program should be required to newly licensed drivers during orientations or licence issuing.

• Journal of IKEEE
• /
• v.22 no.2
• /
• pp.281-288
• /
• 2018

In this paper, we propose an intelligent face recognition and tracking system that distributes GPU resources using CUDA. The proposed system consists of five steps such as GPU allocation algorithm that distributes GPU resources in optimal state, face area detection and face recognition using deep learning, real time face tracking, and PTZ camera control. The GPU allocation algorithm that distributes multi-GPU resources optimally distributes the GPU resources flexibly according to the activation level of the GPU, unlike the method of allocating the GPU to the thread fixedly. Thus, there is a feature that enables stable and efficient use of multiple GPUs. In order to evaluate the performance of the proposed system, we compared the proposed system with the non - distributed system. As a result, the system which did not allocate the resource showed unstable operation, but the proposed system showed stable resource utilization because it was operated stably. Thus, the utility of the proposed system has been demonstrated.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.8
• /
• pp.3567-3582
• /
• 2020

Human activities are often affected by weather conditions. Automatic weather recognition is meaningful to traffic alerting, driving assistance, and intelligent traffic. With the boost of deep learning and AI, deep convolutional neural networks (CNN) are utilized to identify weather situations. In this paper, a three-channel convolutional neural network (3C-CNN) model is proposed on the basis of ResNet50.The model extracts global weather features from the whole image through the ResNet50 branch, and extracts the sky and ground features from the top and bottom regions by two CNN5 branches. Then the global features and the local features are merged by the Concat function. Finally, the weather image is classified by Softmax classifier and the identification result is output. In addition, a medium-scale dataset containing 6,185 outdoor weather images named WeatherDataset-6 is established. 3C-CNN is used to train and test both on the Two-class Weather Images and WeatherDataset-6. The experimental results show that 3C-CNN achieves best on both datasets, with the average recognition accuracy up to 94.35% and 95.81% respectively, which is superior to other classic convolutional neural networks such as AlexNet, VGG16, and ResNet50. It is prospected that our method can also work well for images taken at night with further improvement.