• Journal of Information Processing Systems
• /
• 제19권1호
• /
• pp.130-138
• /
• 2023

The traffic flow in an urban area is affected by the date, weather, and regional traffic flow. The existing methods are weak to model the dynamic road network features, which results in inadequate long-term prediction performance. To solve the problems regarding insufficient capacity for dynamic modeling of road network structures and insufficient mining of dynamic spatio-temporal features. In this study, we propose a novel traffic flow prediction framework called shared spatio-temporal attention convolution optimization network (SSTACON). The shared spatio-temporal attention convolution layer shares a spatio-temporal attention structure, that is designed to extract dynamic spatio-temporal features from historical traffic conditions. Subsequently, the graph optimization module is used to model the dynamic road network structure. The experimental evaluation conducted on two datasets shows that the proposed method outperforms state-of-the-art methods at all time intervals.

• 한국컴퓨터정보학회논문지
• /
• 제24권6호
• /
• pp.21-28
• /
• 2019

Recent years with the development of artificial intelligence and the success of the deep model, they have been deployed in all fields of computer vision. Action recognition, as an important branch of human perception and computer vision system research, has attracted more and more attention. Action recognition is a challenging task due to the special complexity of human movement, the same movement may exist between multiple individuals. The human action exists as a continuous image frame in the video, so action recognition requires more computational power than processing static images. And the simple use of the CNN network cannot achieve the desired results. Recently, the attention model has achieved good results in computer vision and natural language processing. In particular, for video action classification, after adding the attention model, it is more effective to focus on motion features and improve performance. It intuitively explains which part the model attends to when making a particular decision, which is very helpful in real applications. In this paper, we proposed a 3D dense convolutional network based on attention mechanism(ADD-Net), recognition of human motion behavior in the video.

• Computers and Concrete
• /
• 제26권5호
• /
• pp.411-420
• /
• 2020

Deep Convolution neural network (DCNN) has been widely used in the healthy maintenance of civil infrastructure. Using DCNN to improve crack detection performance has attracted many researchers' attention. In this paper, a light-weight spatial attention network module is proposed to strengthen the representation capability of ResNet and improve the crack detection performance. It utilizes attention mechanism to strengthen the interested objects in global receptive field of ResNet convolution layers. Global average spatial information over all channels are used to construct an attention scalar. The scalar is combined with adaptive weighted sigmoid function to activate the output of each channel's feature maps. Salient objects in feature maps are refined by the attention scalar. The proposed spatial attention module is stacked in ResNet50 to detect crack. Experiments results show that the proposed module can got significant performance improvement in crack detection.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권12호
• /
• pp.4439-4455
• /
• 2021

Compared with two-stage object detection algorithms, one-stage algorithms provide a better trade-off between real-time performance and accuracy. However, these methods treat the intermediate features equally, which lacks the flexibility to emphasize meaningful information for classification and location. Besides, they ignore the interaction of contextual information from different scales, which is important for medium and small objects detection. To tackle these problems, we propose an image pyramid network based on dual attention mechanism (DAIPNet), which builds an image pyramid to enrich the spatial information while emphasizing multi-scale informative features based on dual attention mechanisms for one-stage object detection. Our framework utilizes a pre-trained backbone as standard detection network, where the designed image pyramid network (IPN) is used as auxiliary network to provide complementary information. Here, the dual attention mechanism is composed of the adaptive feature fusion module (AFFM) and the progressive attention fusion module (PAFM). AFFM is designed to automatically pay attention to the feature maps with different importance from the backbone and auxiliary network, while PAFM is utilized to adaptively learn the channel attentive information in the context transfer process. Furthermore, in the IPN, we build an image pyramid to extract scale-wise features from downsampled images of different scales, where the features are further fused at different states to enrich scale-wise information and learn more comprehensive feature representations. Experimental results are shown on MS COCO dataset. Our proposed detector with a 300 × 300 input achieves superior performance of 32.6% mAP on the MS COCO test-dev compared with state-of-the-art methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권2호
• /
• pp.544-564
• /
• 2022

Underwater images often suffer from color distortion, blurring and low contrast, which is caused by the propagation of light in the underwater environment being affected by the two processes: absorption and scattering. To cope with the poor quality of underwater images, this paper proposes a multiscale fusion underwater image enhancement method based on channel attention mechanism and local binary pattern (LBP). The network consists of three modules: feature aggregation, image reconstruction and LBP enhancement. The feature aggregation module aggregates feature information at different scales of the image, and the image reconstruction module restores the output features to high-quality underwater images. The network also introduces channel attention mechanism to make the network pay more attention to the channels containing important information. The detail information is protected by real-time superposition with feature information. Experimental results demonstrate that the method in this paper produces results with correct colors and complete details, and outperforms existing methods in quantitative metrics.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2023년도 춘계학술발표대회
• /
• pp.493-495
• /
• 2023

Deep learning methods and attention mechanisms have been incorporated to improve facial emotion recognition, which has recently attracted much attention. The fusion approaches have improved accuracy by combining various types of information. This research proposes a fusion network with self-attention and local attention mechanisms. It uses a multi-layer perceptron network. The network extracts distinguishing characteristics from facial images using pre-trained models on RAF-DB dataset. We outperform the other fusion methods on RAD-DB dataset with impressive results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권11호
• /
• pp.4426-4442
• /
• 2020

Since scene segmentation is becoming a hot topic in the field of autonomous driving and medical image analysis, researchers are actively trying new methods to improve segmentation accuracy. At present, the main issues in image semantic segmentation are intra-class inconsistency and inter-class indistinction. From our analysis, the lack of global information as well as macroscopic discrimination on the object are the two main reasons. In this paper, we propose a Densely connected residual Attention network (DA-Res2Net) which consists of a dense residual network and channel attention guidance module to deal with these problems and improve the accuracy of image segmentation. Specifically, in order to make the extracted features equipped with stronger multi-scale characteristics, a densely connected residual network is proposed as a feature extractor. Furthermore, to improve the representativeness of each channel feature, we design a Channel-Attention-Guide module to make the model focusing on the high-level semantic features and low-level location features simultaneously. Experimental results show that the method achieves significant performance on various datasets. Compared to other state-of-the-art methods, the proposed method reaches the mean IOU accuracy of 83.2% on PASCAL VOC 2012 and 79.7% on Cityscapes dataset, respectively.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2020년도 제32회 한글 및 한국어 정보처리 학술대회
• /
• pp.333-336
• /
• 2020

개체명 인식이란 문장에서 인명, 지명, 기관명 등과 같이 고유한 의미를 갖는 단어를 찾아 개체명을 분류하는 작업이다. 딥러닝을 활용한 연구가 수행되면서 개체명 인식에 RNN(Recurrent Neural Network)과 CRF(Condition Random Fields)를 결합한 연구가 좋은 성능을 보이고 있다. 그러나 CRF는 시간 복잡도가 분류해야 하는 클래스(Class) 개수의 제곱에 비례하고, 최근 RNN과 Softmax 모델보다 낮은 성능을 보이는 연구도 있었다. 본 논문에서는 CRF의 단점을 보완한 LAN(Label Attention Network)와 사전 학습 언어 모델인 음절 단위 ELECTRA를 활용하는 개체명 인식 모델을 제안한다.

• 전기전자학회논문지
• /
• 제27권3호
• /
• pp.251-257
• /
• 2023

본 논문에서는 비전 트랜스포머의 셀프 어텐션이 갖는 지역적 특징 부족을 개선하는 이중 구조 셀프 어텐션 방법을 제안한다. 객체 분류, 객체 분할, 비디오 영상 인식에서 합성곱 신경망보다 연산 효율성이 높은 비전 트랜스포머는 상대적으로 지역적 특징 추출능력이 부족하다. 이 문제를 해결하기 위해 윈도우 또는 쉬프트 윈도우를 기반으로 하는 연구가 많이 이루어지고 있으나 이러한 방법은 여러 단계의 인코더를 사용하여 연산 복잡도의 증가로 셀프 어텐션 기반 트랜스포머의 장점이 약화 된다. 본 논문에서는 기존의 방법보다 locality inductive bias 향상을 위해 self-attention과 neighborhood network를 이용하여 이중 구조 셀프 어텐션을 제안한다. 지역적 컨텍스트 정보 추출을 위한 neighborhood network은 윈도우 구조보다 훨씬 단순한 연산 복잡도를 제공한다. 제안된 이중 구조 셀프 어텐션 트랜스포머와 기존의 트랜스포머의 성능 비교를 위해 CIFAR-10과 CIFAR-100을 학습 데이터를 사용하였으며 실험결과 Top-1 정확도에서 각각 0.63%과 1.57% 성능이 개선되었다.

• 한국컴퓨터정보학회논문지
• /
• 제25권1호
• /
• pp.55-61
• /
• 2020

딥 러닝 기술의 발전과 컴퓨팅 파워 등의 개선으로 인해 비디오 기반 연구는 최근 많은 관심을 얻고 있다. 비디오 데이터가 이미지 데이터와 비교하여 가장 큰 차이는 비디오 데이터에는 많은 양의 시간적, 공간적 정보가 포함되어 있다는 점이다. 이처럼 비디오에 포함된 많은 양의 데이터로 인해 컴퓨터 비전 연구에 있어서 행동 인식은 중요한 연구 과제 중 하나이지만, 비디오와 같이 움직임이 있는 환경에서 인간의 행동 인식은 매우 복잡하고 도전적인 과제이다. 인간에 대한 여러 연구를 바탕으로 인공지능에서는 인간과 유사한 주의(attention)메커니즘이 효율적인 인식 모델이라는 것을 알게 되었다. 이 효율적인 모델은 이미지 정보와 복잡한 연속 비디오 정보를 처리하는 데 이상적이다. 본 논문에서는 이러한 연구배경을 기반으로, 비디오에서 인간의 행동을 효율적으로 인식하기 위해 먼저 인간의 행동에 주목한 후 비디오 행동 인식에 주의메커니즘을 도입하고자 한다. 논문의 주요내용은 두 가지 주의 메카니즘을 기반으로 컨볼루션 신경망을 이용한 새로운 3D 잔류 주의 네트워크를 제안함으로써 비디오에서 인간의 행동을 식별하고자 한다. 제안 모델의 평가 결과 최대 90.7%정도의 정확도를 보였다.