• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.3
• /
• pp.401-406
• /
• 2021

Recently, mean-teacher models based on convolutional recurrent neural networks are popularly used in audio event detection. The mean-teacher model is an architecture that consists of two parallel CRNNs and it is possible to train them effectively on the weakly-labelled and unlabeled audio data by using the consistency learning metric at the output of the two neural networks. In this study, we tried to improve the performance of the mean-teacher model by using additional derivative features of the log-mel spectrum. In the audio event detection experiments using the training and test data from the Task 4 of the DCASE 2018/2019 Challenges, we could obtain maximally a 8.1% relative decrease in the ER(Error Rate) in the mean-teacher model using proposed derivative features.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.5
• /
• pp.17-28
• /
• 2023

Potholes that occur on paved roads can have fatal consequences for vehicles traveling at high speeds and may even lead to fatalities. While manual detection of potholes using human labor is commonly used to prevent pothole-related accidents, it is economically and temporally inefficient due to the exposure of workers on the road and the difficulty in predicting potholes in certain categories. Therefore, completely preventing potholes is nearly impossible, and even preventing their formation is limited due to the influence of ground conditions closely related to road environments. Additionally, labeling work guided by experts is required for dataset construction. Thus, in this paper, we utilized the Mean Teacher technique, one of the semi-supervised learning-based knowledge distillation methods, to achieve robust performance in pothole image classification even with limited labeled data. We demonstrated this using performance metrics and GradCAM, showing that when using semi-supervised learning, 15 pre-trained CNN models achieved an average accuracy of 90.41%, with a minimum of 2% and a maximum of 9% performance difference compared to supervised learning.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.5
• /
• pp.395-401
• /
• 2023

Sound Event Detection (SED) aims to identify not only sound category but also time interval for target sounds in an audio waveform. It is a critical technique in field of acoustic surveillance system and monitoring system. Recently, various models have introduced through Detection and Classification of Acoustic Scenes and Events (DCASE) Task 4. This paper explored how to design optimal parameters of DenseNet based model, which has led to outstanding performance in other recognition system. In experiment, DenseRNN as an SED model consists of DensNet-BC and bi-directional Gated Recurrent Units (GRU). This model is trained with Mean teacher model. With an event-based f-score, evaluation is performed depending on parameters, related to model architecture as well as model training, under the assessment protocol of DCASE task4. Experimental result shows that the performance goes up and has been saturated to near the best. Also, DenseRNN would be trained more effectively without dropout technique.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.139-147
• /
• 2021

This paper introduces and implements a Sound Event Detection (SED) system based on weakly-supervised learning where only part of the data is labeled, and analyzes the effect of parameters. The SED system estimates the classes and onset/offset times of events in the acoustic signal. In order to train the model, all information on the event class and onset/offset times must be provided. Unfortunately, the onset/offset times are hard to be labeled exactly. Therefore, in the weakly-supervised task, the SED model is trained by "strongly labeled data" including the event class and activations, "weakly labeled data" including the event class, and "unlabeled data" without any label. Recently, the SED systems using the mean-teacher model are widely used for the task with several parameters. These parameters should be chosen carefully because they may affect the performance. In this paper, performance analysis was performed on parameters, such as the feature, moving average parameter, weight of the consistency cost function, ramp-up length, and maximum learning rate, using the data of DCASE 2020 Task 4. Effects and the optimal values of the parameters were discussed.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.27 no.5
• /
• pp.113-119
• /
• 2023

Most infrastructure structures were completed during periods of economic growth. The number of infrastructure structures reaching their lifespan is increasing, and the proportion of old structures is gradually increasing. The functions and performance of these structures at the time of design may deteriorate and may even lead to safety accidents. To prevent this repercussion, accurate inspection and appropriate repair are requisite. To this end, demand is increasing for computer vision and deep learning technology to accurately detect even minute cracks. However, deep learning algorithms require a large number of training data. In particular, label images indicating the location of cracks in the image are required. To secure a large number of those label images, a lot of labor and time are consumed. To reduce these costs as well as increase detection accuracy, this study proposed a learning structure based on mean teacher method. This learning structure was trained on a dataset of 900 labeled image dataset and 3000 unlabeled image dataset. The crack detection network model was evaluated on over 300 labeled image dataset, and the detection accuracy recorded a mean intersection over union of 89.23% and an F1 score of 89.12%. Through this experiment, it was confirmed that detection performance was improved compared to supervised learning. It is expected that this proposed method will be used in the future to reduce the cost required to secure label images.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.5
• /
• pp.21-30
• /
• 2023

Accurate segmentation of the kidney tumor is necessary to identify shape, location and safety margin of tumor in abdominal CT images for surgical planning before renal partial nephrectomy. However, kidney tumor segmentation is challenging task due to the various sizes and locations of the tumor for each patient and signal intensity similarity to surrounding organs such as intestine and spleen. In this paper, we propose a semi-supervised learning-based mean teacher network that utilizes both labeled and unlabeled data using a kidney local guided map including kidney local information to segment small-sized kidney tumors occurring at various locations in the kidney, and analyze the performance according to the kidney tumor size. As a result of the study, the proposed method showed an F1-score of 75.24% by considering local information of the kidney using a kidney local guide map to locate the tumor existing around the kidney. In particular, under-segmentation of small-sized tumors which are difficult to segment was improved, and showed a 13.9%p higher F1-score even though it used a smaller amount of labeled data than nnU-Net.

• Journal of The Korean Association of Information Education
• /
• v.22 no.3
• /
• pp.307-316
• /
• 2018

In this study, we conduct SW education using robot in elementary school, to examine the effects of SW education using robot and identify differences dependeing on the presence of assistance teacher. In order to achieve the purpose of this study, SW education using robot was conducted to 29 students. After collecting data, we examined mean difference using matched pair t-test and ANCOVA. Our results show that SW education using robot significantly improved CT and attitudes toward robots. CT, however, was not significantly improved in control group. Further, although improvement in attitudes toward robots was not associated with presence of assistance teacher, presence of assistance teacher did play a role in improvement of CT. This study was investigated effects of the SW education using robot and identify differences depending on the presence of assistant teacher, and expanded the understanding of the SW education using robot.

• The Journal of Korean Association of Computer Education
• /
• v.3 no.1
• /
• pp.127-134
• /
• 2000

Distance education, which emerged under the influence of both the rapid development of hi-tech information technology and the constructivism theory of learning, enables learners to acquire knowledge and skills needed by monitoring their learning process for themselves. Emphasis on web-based distance education and contructivism as a basis of learner-centered education does not mean that those education systems are performed without teachers or in noninterference. This study proposes a model of distance education system in which learners are taught in various levels of learning with the help of teacher agents. In the model teacher agents produce the learners' respective learning model considering the information on individual learners and also control the progress to the next step of learning with the result of the evaluation of learning accomplishment. The model of distance education system suggested in this study is able to help solve the problem that the learning resources web-based distance education provides were used only as objects of web search and to supply a basis of realizing learner-centered education.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.2
• /
• pp.51-58
• /
• 2023

This paper proposes a deep learning method for estimating the heart rate from facial videos. Our proposed method estimates remote photoplethysmography (rPPG) signals to predict the heart rate. Although there have been proposed several methods for estimating rPPG signals, most previous methods can not be utilized in low-power single board computers due to their computational complexity. To address this problem, we construct a lightweight student model and employ a knowledge distillation technique to reduce the performance degradation of a deeper network model. The teacher model consists of 795k parameters, whereas the student model only contains 24k parameters, and therefore, the inference time was reduced with the factor of 10. By distilling the knowledge of the intermediate feature maps of the teacher model, we improved the accuracy of the student model for estimating the heart rate. Experiments were conducted on the UBFC-rPPG dataset to demonstrate the effectiveness of the proposed method. Moreover, we collected our own dataset to verify the accuracy and processing time of the proposed method on a real-world dataset. Experimental results on a NVIDIA Jetson Nano board demonstrate that our proposed method can infer the heart rate in real time with the mean absolute error of 2.5183 bpm.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.479-487
• /
• 2021

In this paper, we propose an Sound Event Detection (SED) model using self-training based on a noisy student model. The proposed SED model consists of two stages. In the first stage, a mean-teacher model based on an Residual Convolutional Recurrent Neural Network (RCRNN) is constructed to provide target labels regarding weakly labeled or unlabeled data. In the second stage, a self-training-based noisy student model is constructed by applying different noise types. That is, feature noises, such as time-frequency shift, mixup, SpecAugment, and dropout-based model noise are used here. In addition, a semi-supervised loss function is applied to train the noisy student model, which acts as label noise injection. The performance of the proposed SED model is evaluated on the validation set of the Detection and Classification of Acoustic Scenes and Events (DCASE) 2020 Challenge Task 4. The experiments show that the single model and ensemble model of the proposed SED based on the noisy student model improve F1-score by 4.6 % and 3.4 % compared to the top-ranked model in DCASE 2020 challenge Task 4, respectively.