• ETRI Journal
• /
• 제39권5호
• /
• pp.643-651
• /
• 2017

Human age estimation is one of the key factors in the field of Human-Robot Interaction/Human-Computer Interaction (HRI/HCI). Owing to the development of deep-learning technologies, age recognition has recently been attempted. In general, however, deep learning techniques require a large-scale database, and for age learning with variations, a conventional database is insufficient. For this reason, we propose an age estimation method using artificially generated data. Image data are artificially generated through 3D information, thus solving the problem of shortage of training data, and helping with the training of the deep-learning technique. Augmentation using 3D has advantages over 2D because it creates new images with more information. We use a deep architecture as a pre-trained model, and improve the estimation capacity using artificially augmented training images. The deep architecture can outperform traditional estimation methods, and the improved method showed increased reliability. We have achieved state-of-the-art performance using the proposed method in the Morph-II dataset and have proven that the proposed method can be used effectively using the Adience dataset.

• 한국전자통신학회논문지
• /
• 제18권2호
• /
• pp.329-335
• /
• 2023

타임지가 선정한 슈퍼푸드이며, 후숙 과일 중 하나인 아보카도는 현지가격과 국내 유통 가격이 크게 차이가 나는 식품 중 하나이다. 이러한 아보카도의 분류과정을 자동화한다면 다양한 분야에서 인건비를 줄여 가격을 낮출 수 있을 것이다. 본 논문에서는 아보카도의 데이터셋을 크롤링을 통하여 제작하고, 딥러닝 기반 전이학습모델을 다수 사용하여, 최적의 분류모델을 만드는 것을 목표로 한다. 실험은 제작한 데이터셋에서 분리한 데이터셋에서 딥러닝 기반 전이학습모델에 직접 대입하고, 해당 모델의 하이퍼 파라미터를 Fine-tuning하며 진행하였다. 제작된 모델은 아보카도의 이미지를 입력하였을 때, 해당 아보카도의 익은 정도를 99% 이상의 정확도로 분류하였으며, 아보카도 생산 및 유통가정의 인력감소 및 정확성을 높일 수 있는 데이터셋 및 알고리즘을 제안한다.

• 생물환경조절학회지
• /
• 제31권3호
• /
• pp.255-260
• /
• 2022

본 논문은 딥러닝 알고리즘을 이용하여 딸기 영상 데이터의 병충해 존재 여부를 자동으로 검출할 수 있는 서비스 모델을 제안한다. 또한 병징에 특화된 분할 이미지 데이터 세트를 제안하여 딥러닝 모델의 병충해 검출 성능을 향상한다. 딥러닝 모델은 CNN 기반 YOLO를 선정하여 기존의 R-CNN 기반 모델의 느린 학습속도와 추론속도를 개선하였다. 병충해 검출 모델을 학습하기 위해 일반적인 데이터 세트와 제안하는 분할 이미지 데이터 세트를 구축하였다. 딥러닝 모델이 일반적인 학습 데이터 세트를 학습했을 때 병충해 검출률은 81.35%이며 병충해 검출 신뢰도는 73.35%이다. 반면 딥러닝 모델이 분할 이미지 학습 데이터 세트를 학습했을 때 병충해 검출률은 91.93%이며 병충해 검출 신뢰도는 83.41%이다. 따라서 분할 이미지 데이터를 학습한 딥러닝 모델의 성능이 우수하다는 것을 증명할 수 있었다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권11호
• /
• pp.3658-3679
• /
• 2022

Classification of persons wearing and not wearing face masks in images has emerged as a new computer vision problem during the COVID-19 pandemic. In order to address this problem and scale up the research in this domain, in this paper a hybrid technique by employing ResNet-101 and multi-layer perceptron (MLP) classifier has been proposed. The proposed technique is tested and validated on a self-created face masks classification dataset and a standard dataset. On self-created dataset, the proposed technique achieved a classification accuracy of 97.3%. To embrace the proposed technique, six other state-of-the-art CNN feature extractors with six other classical machine learning classifiers have been tested and compared with the proposed technique. The proposed technique achieved better classification accuracy and 1-6% higher precision, recall, and F1 score as compared to other tested deep feature extractors and machine learning classifiers.

• 대한원격탐사학회지
• /
• 제39권1호
• /
• pp.87-97
• /
• 2023

Unmanned aerial vehicles (UAVs) can capture high-resolution imagery from a variety of viewing angles and altitudes; they are generally limited to collecting images of small scenes from larger regions. To improve the utility of UAV-appropriated datasetsfor use with deep learning applications, multiple datasets created from variousregions under different conditions are needed. To demonstrate a powerful new method for integrating heterogeneous UAV datasets, this paper applies a combined segmentation network (CSN) to share UAVid and semantic drone dataset encoding blocks to learn their general features, whereas its decoding blocks are trained separately on each dataset. Experimental results show that our CSN improves the accuracy of specific classes (e.g., cars), which currently comprise a low ratio in both datasets. From this result, it is expected that the range of UAV dataset utilization will increase.

• 대한의용생체공학회:의공학회지
• /
• 제44권5호
• /
• pp.303-314
• /
• 2023

This study aimed to assess the practical feasibility of advanced deformable image registration (DIR) algorithms in radiotherapy by employing two distinct datasets. The first dataset included 14 4D lung CT scans and 31 head and neck CT scans. In the 4D lung CT dataset, we employed the DIR algorithm to register organs at risk and tumors based on respiratory phases. The second dataset comprised pre-, mid-, and post-treatment CT images of the head and neck region, along with organ at risk and tumor delineations. These images underwent registration using the DIR algorithm, and Dice similarity coefficients (DSCs) were compared. In the 4D lung CT dataset, registration accuracy was evaluated for the spinal cord, lung, lung nodules, esophagus, and tumors. The average DSCs for the non-learning-based SyN and NiftyReg algorithms were 0.92±0.07 and 0.88±0.09, respectively. Deep learning methods, namely Voxelmorph, Cyclemorph, and Transmorph, achieved average DSCs of 0.90±0.07, 0.91±0.04, and 0.89±0.05, respectively. For the head and neck CT dataset, the average DSCs for SyN and NiftyReg were 0.82±0.04 and 0.79±0.05, respectively, while Voxelmorph, Cyclemorph, and Transmorph showed average DSCs of 0.80±0.08, 0.78±0.11, and 0.78±0.09, respectively. Additionally, the deep learning DIR algorithms demonstrated faster transformation times compared to other models, including commercial and conventional mathematical algorithms (Voxelmorph: 0.36 sec/images, Cyclemorph: 0.3 sec/images, Transmorph: 5.1 sec/images, SyN: 140 sec/images, NiftyReg: 40.2 sec/images). In conclusion, this study highlights the varying clinical applicability of deep learning-based DIR methods in different anatomical regions. While challenges were encountered in head and neck CT registrations, 4D lung CT registrations exhibited favorable results, indicating the potential for clinical implementation. Further research and development in DIR algorithms tailored to specific anatomical regions are warranted to improve the overall clinical utility of these methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권2호
• /
• pp.632-657
• /
• 2022

Detecting web attacks is a major challenge, and it is observed that the use of simple models leads to low sensitivity or high false positive problems. In this study, we aim to develop a robust two-stage deep learning based stacked ensemble web application firewall. Normal and abnormal classification is carried out in the first stage of the proposed WAF model. The classification process of the types of abnormal traffics is postponed to the second stage and carried out using an integrated stacked ensemble model. By this way, clients' requests can be served without time delay, and attack types can be detected with high sensitivity. In addition to the high accuracy of the proposed model, by using the statistical similarity and diversity analyses in the study, high generalization for the ensemble model is achieved. Within the study, a comprehensive, up-to-date, and robust multi-class web anomaly dataset named GAZI-HTTP is created in accordance with the real-world situations. The performance of the proposed WAF model is compared to state-of-the-art deep learning models and previous studies using the benchmark dataset. The proposed two-stage model achieved multi-class detection rates of 97.43% and 94.77% for GAZI-HTTP and ECML-PKDD, respectively.

• 산업경영시스템학회지
• /
• 제47권1호
• /
• pp.9-19
• /
• 2024

Deep learning-based computer vision anomaly detection algorithms are widely utilized in various fields. Especially in the manufacturing industry, the difficulty in collecting abnormal data compared to normal data, and the challenge of defining all potential abnormalities in advance, have led to an increasing demand for unsupervised learning methods that rely on normal data. In this study, we conducted a comparative analysis of deep learning-based unsupervised learning algorithms that define and detect abnormalities that can occur when transparent contact lenses are immersed in liquid solution. We validated and applied the unsupervised learning algorithms used in this study to the existing anomaly detection benchmark dataset, MvTecAD. The existing anomaly detection benchmark dataset primarily consists of solid objects, whereas in our study, we compared unsupervised learning-based algorithms in experiments judging the shape and presence of lenses submerged in liquid. Among the algorithms analyzed, EfficientAD showed an AUROC and F1-score of 0.97 in image-level tests. However, the F1-score decreased to 0.18 in pixel-level tests, making it challenging to determine the locations where abnormalities occurred. Despite this, EfficientAD demonstrated excellent performance in image-level tests classifying normal and abnormal instances, suggesting that with the collection and training of large-scale data in real industrial settings, it is expected to exhibit even better performance.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2019년도 춘계학술대회
• /
• pp.347-348
• /
• 2019

본 논문에서는 딥러닝을 사용하여 저해상도의 디지털 홀로그램을 고해상도로 변환하는 방법을 제안한다. 홀로그램의 크기를 축소한 뒤, 축소된 홀로그램을 분할하여 데이터 세트를 생성하였다. 생성한 데이터 세트를 이용하여 네트워크 모델을 훈련시켰으며, 31dB 이상의 결과를 확인할 수 있었다.

• Journal of information and communication convergence engineering
• /
• 제21권4호
• /
• pp.294-299
• /
• 2023

Micro ribonucleic acids (miRNAs) can regulate the protein expression levels of genes in the human body and have recently been reported to be closely related to the cause of disease. Determining the genes related to miRNAs will aid in understanding the mechanisms underlying complex miRNAs. However, the identification of miRNA-related genes through wet experiments (in vivo, traditional methods are time- and cost-consuming). To overcome these problems, recent studies have investigated the prediction of miRNA relevance using deep learning models. This study presents a method for predicting the relationships between miRNAs and genes. First, we reconstruct a negative dataset using the proposed method. We then extracted the feature using an autoencoder, after which the feature vector was concatenated with the original data. Thereafter, the concatenated data were used to train a long short-term memory model. Our model exhibited an area under the curve of 0.9609, outperforming previously reported models trained using the same dataset.