• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.11
• /
• pp.3493-3506
• /
• 2022

In the field of scene segmentation, the precise segmentation of object boundaries in sports movement scene images is a great challenge. The geometric information and spatial information of the image are very important, but in many models, they are usually easy to be lost, which has a big influence on the performance of the model. To alleviate this problem, a parallel dense dilated convolution merging Network (termed PDDCM-Net) was proposed. The proposed PDDCMNet consists of a feature extractor, parallel dilated convolutions, and dense dilated convolutions merged with different dilation rates. We utilize different combinations of dilated convolutions that expand the receptive field of the model with fewer parameters than other advanced methods. Importantly, PDDCM-Net fuses both low-level and high-level information, in effect alleviating the problem of accurately segmenting the edge of the object and positioning the object position accurately. Experimental results validate that the proposed PDDCM-Net achieves a great improvement compared to several representative models on the COCO-Stuff data set.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.414-423
• /
• 2020

In this paper, we propose a Dilated Convolution Gate Linear Unit (DCGLU) to mitigate the lack of sparsity and small receptive field problems caused by the segmentation map extraction process in sound event detection with weak labels. In the advent of deep learning framework, segmentation map extraction approaches have shown improved performance in noisy environments. However, these methods are forced to maintain the size of the feature map to extract the segmentation map as the model would be constructed without a pooling operation. As a result, the performance of these methods is deteriorated with a lack of sparsity and a small receptive field. To mitigate these problems, we utilize GLU to control the flow of information and Dilated Convolutional Neural Networks (DCNNs) to increase the receptive field without additional learning parameters. For the performance evaluation, we employ a URBAN-SED and self-organized bird sound dataset. The relevant experiments show that our proposed DCGLU model outperforms over other baselines. In particular, our method is shown to exhibit robustness against nature sound noises with three Signal to Noise Ratio (SNR) levels (20 dB, 10 dB and 0 dB).

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.9
• /
• pp.2458-2482
• /
• 2023

Medical image segmentation techniques based on convolution neural networks indulge in feature extraction triggering redundancy of parameters and unsatisfactory target localization, which outcomes in less accurate segmentation results to assist doctors in diagnosis. In this paper, we propose a multi-level semantic-rich encoding-decoding network, which consists of a Pooling-Conv-Former (PCFormer) module and a Cbam-Dilated-Transformer (CDT) module. In the PCFormer module, it is used to tackle the issue of parameter explosion in the conservative transformer and to compensate for the feature loss in the down-sampling process. In the CDT module, the Cbam attention module is adopted to highlight the feature regions by blending the intersection of attention mechanisms implicitly, and the Dilated convolution-Concat (DCC) module is designed as a parallel concatenation of multiple atrous convolution blocks to display the expanded perceptual field explicitly. In addition, MultiHead Attention-DwConv-Transformer (MDTransformer) module is utilized to evidently distinguish the target region from the background region. Extensive experiments on medical image segmentation from Glas, SIIM-ACR, ISIC and LGG demonstrated that our proposed network outperforms existing advanced methods in terms of both objective evaluation and subjective visual performance.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.11a
• /
• pp.84-86
• /
• 2018

요즘 자율주행과 같은 최신 기술의 발전과 더불어 촬영된 영상 장면에 대한 깊이있는 이해가 필요하게 되었다. 특히, 기계학습 기술이 발전하면서 카메라로 찍은 영상에 대한 의미론적 분할 기술에 대한 연구도 활발히 진행되고 있다. FuseNet은 인코더-디코더 구조를 이용하여 장면 내에 있는 객체에 대한 의미론적 분할 기술을 적용할 수 있는 신경망 모델이다. FuseNet은 오직 RGB 입력을 받는 기존의 FCN보다 깊이정보까지 활용하여 RGB 정보를 기반으로 추출한 특징지도와의 요소합 연산을 통해 멀티 모달 구조를 구현했다. 의미론적 분할 연구에서는 객체의 전역 컨텍스트가 고려되는 것이 중요한데, 이를 위해 여러 계층을 깊게 쌓으면 연산량이 많아지는 단점이 있다. 이를 극복하기 위해서 기존의 합성곱 방식을 벗어나 새롭게 제안된 팽창 합성곱 연산(Dilated Convolution)을 이용하면 객체의 수용 영역이 효과적으로 넓어지고 연산량이 적어질 수 있다. 본 논문에서는 컨볼루션 연산의 새로운 방법론적 접근 중 하나인 팽창된 합성곱 연산을 이용해 의미론적 분할 연구에서 새로운 멀티 모달 네트워크의 성능 향상 방법을 적용하여 계층을 더 깊게 쌓지 않더라도 파라미터의 증가 없이 해상도를 유지하면서 네트워크의 전체 성능을 향상할 수 있는 최적화된 방법을 제안한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1778-1797
• /
• 2021

Document image binarization is an important pre-processing step in document analysis and archiving. The state-of-the-art models for document image binarization are variants of encoder-decoder architectures, such as FCN (fully convolutional network) and U-Net. Despite their success, they still suffer from three limitations: (1) reduced feature map resolution due to consecutive strided pooling or convolutions, (2) multiple scales of target objects, and (3) reduced localization accuracy due to the built-in invariance of deep convolutional neural networks (DCNNs). To overcome these three challenges, we propose an improved semantic segmentation model, referred to as DP-LinkNet, which adopts the D-LinkNet architecture as its backbone, with the proposed hybrid dilated convolution (HDC) and spatial pyramid pooling (SPP) modules between the encoder and the decoder. Extensive experiments are conducted on recent document image binarization competition (DIBCO) and handwritten document image binarization competition (H-DIBCO) benchmark datasets. Results show that our proposed DP-LinkNet outperforms other state-of-the-art techniques by a large margin. Our implementation and the pre-trained models are available at https://github.com/beargolden/DP-LinkNet.

• Journal of the Korean Society of Industry Convergence
• /
• v.26 no.2_1
• /
• pp.217-223
• /
• 2023

With the rapid advancement of technologies, need for different research fields where this technology can be used is also increasing. One of the most researched topic in computer vision is object detection, which has widely been implemented in various fields which include healthcare, video surveillance and education. The main goal of object detection is to identify and categorize all the objects in a target environment. Specifically, methods of object detection consist of a variety of significant techniq ues, such as image processing and patterns recognition. Anomaly detection is a part of object detection, anomalies can be found various scenarios for example crowded places such as subway stations. An abnormal event can be assumed as a variation from the conventional scene. Since the abnormal event does not occur frequently, the distribution of normal and abnormal events is thoroughly imbalanced. In terms of public safety, abnormal events should be avoided and therefore immediate action need to be taken. When abnormal events occur in certain places, real time detection is required to prevent and protect the safety of the people. To solve the above problems, we propose a modified YOLOv5 object detection algorithm by implementing dilated convolutional layers which achieved 97% mAP50 compared to other five different models of YOLOv5. In addition to this, we also created a simple mobile application to avail the abnormal event detection on mobile phones.

• Journal of Korea Multimedia Society
• /
• v.23 no.9
• /
• pp.1129-1138
• /
• 2020

The performance of convolutional deep learning networks is generally determined according to parameters of target dataset, structure of network, convolution kernel, activation function, and optimization algorithm. In this paper, a genetic algorithm is used to select the appropriate deep learning model and parameters for Alzheimer's classification and to compare the learning results with preliminary experiment. We compare and analyze the Alzheimer's disease classification performance of VGG-16, GoogLeNet, and ResNet to select an effective network for detecting AD and MCI. The simulation results show that the network structure is ResNet, the activation function is ReLU, the optimization algorithm is Adam, and the convolution kernel has a 3-dilated convolution filter for the accuracy of dementia medical images.

• Journal of the Korea Convergence Society
• /
• v.12 no.12
• /
• pp.31-37
• /
• 2021

In this paper proposed a feature fusion and spatial attention-based modified YOLOv4S for small and occluded detection. Conventional YOLOv4S is a lightweight network and lacks feature extraction capability compared to the method of the deep network. The proposed method first combines feature maps of different scales with feature fusion to enhance semantic and low-level information. In addition expanding the receptive field with dilated convolution, the detection accuracy for small and occluded objects was improved. Second by improving the conventional spatial information with spatial attention, the detection accuracy of objects classified and occluded between objects was improved. PASCAL VOC and COCO datasets were used for quantitative evaluation of the proposed method. The proposed method improved mAP by 2.7% in the PASCAL VOC dataset and 1.8% in the COCO dataset compared to the Conventional YOLOv4S.

• Journal of the Korea Society for Simulation
• /
• v.30 no.4
• /
• pp.1-8
• /
• 2021

Deep learning is one of the most widely accepted methods for the forecasting of time series data which have the complexity and non-linear behavior. In this paper, we investigate the modification of a state-of-art WaveNet deep learning architecture and walk forward validation (WFV) in order to forecast electric power consumption data 24-hour-ahead. WaveNet originally designed for raw audio uses 1D dilated causal convolution for long-term information. First of all, we propose a modified version of WaveNet which activates real numbers instead of coded integers. Second, this paper provides with the training process with tuning of major hyper-parameters (i.e., input length, batch size, number of WaveNet blocks, dilation rates, and learning rate scheduler). Finally, performance evaluation results show that the prediction methodology based on WFV performs better than on the traditional holdout validation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.9
• /
• pp.3583-3597
• /
• 2020

In this paper, a parallel deep learning model using a convolutional neural network and a dilated convolutional neural network is proposed to classify Alzheimer's disease with high accuracy in PET/CT images. The developed model consists of two pipelines, a conventional CNN pipeline, and a dilated convolution pipeline. An input image is sent through both pipelines, and at the end of both pipelines, extracted features are concatenated and used for classifying Alzheimer's disease. Complimentary abilities of both networks provide better overall accuracy than single conventional CNNs in the dataset. Moreover, instead of performing binary classification, the proposed model performs three-class classification being Alzheimer's disease, mild cognitive impairment, and normal control. Using the data received from Dong-a University, the model performs classification detecting Alzheimer's disease with an accuracy of up to 95.51%.