• Journal of Digital Contents Society
• /
• v.19 no.6
• /
• pp.1213-1218
• /
• 2018

Recently introduced Mask R - CNN presents a conceptually simple, flexible, general framework for instance segmentation of objects. In this paper, we propose an algorithm for efficiently searching objects of images, while creating a segmentation mask of heat generation part for an instance which is a heating element in a heat sensed image acquired from a thermal infrared camera. This method called a mask R - CNN is an algorithm that extends Faster R - CNN by adding a branch for predicting an object mask in parallel with an existing branch for recognition of a bounding box. The mask R - CNN is added to the high - speed R - CNN which training is easy and fast to execute. Also, it is easy to generalize the mask R - CNN to other tasks. In this research, we propose an infrared image detection algorithm based on R - CNN and detect heating elements which can not be distinguished by RGB images. As a result of the experiment, a heat-generating object which can not be discriminated from Mask R-CNN was detected normally.

• The Journal of Korean Institute of Information Technology
• /
• v.16 no.11
• /
• pp.105-113
• /
• 2018

In this paper, we proposed a method for effective classification of eye, nose, and mouth of human face. Most recent image classification uses Convolutional Neural Network(CNN). However, the features extracted by CNN are not sufficient and the classification effect is not too high. We proposed a new algorithm to improve the classification effect. The proposed method can be roughly divided into three parts. First, the Haar feature extraction algorithm is used to construct the eye, nose, and mouth dataset of face. The second, the model extracts CNN features of image using AlexNet. Finally, Haar-CNN features are extracted by performing convolution after Haar feature extraction. After that, CNN features and Haar-CNN features are fused and classify images using softmax. Recognition rate using mixed features could be increased about 4% than CNN feature. Experiments have demonstrated the performance of the proposed algorithm.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.181-187
• /
• 2024

There are two kinds of well-known CNN methods, the group equivariant CNN and the CNN using steerable filters, which have excellent classification performances for randomly rotated objects in image space. This paper describes their mathematical structures and introduces implementation methods. We implement them, including the existing CNN, which have the same number of filters, then compare and analyze their performances by simulating them with the randomly rotated MNIST. According to the experimental results, the steerable CNN, which shows a classification improvement over the others, has a relatively small number of parameters to learn, so performance degradation is relatively small even when the size of the training dataset is reduced.

• Journal of IKEEE
• /
• v.25 no.1
• /
• pp.180-192
• /
• 2021

One widely used algorithm for image recognition and pattern detection is the convolution neural network (CNN). To efficiently handle convolution operations, which account for the majority of computations in the CNN, we use hardware accelerators to improve the performance of CNN applications. In using these hardware accelerators, the CNN fetches data from the off-chip DRAM, as the massive computational volume of data makes it difficult to derive performance improvements only from memory inside the hardware accelerator. In other words, data communication between off-chip DRAM and memory inside the accelerator has a significant impact on the performance of CNN applications. In this paper, a simulator for the CNN is developed to analyze the main memory or DRAM access with respect to the size of the on-chip memory or global buffer inside the CNN accelerator. For AlexNet, one of the CNN architectures, when simulated with increasing the size of the global buffer, we found that the global buffer of size larger than 100kB has 0.8x as low a DRAM access count as the global buffer of size smaller than 100kB.

• Journal of Internet Computing and Services
• /
• v.9 no.3
• /
• pp.1-8
• /
• 2008

With growing interest in location-based service (LBS), there is increasing necessity for nearest neighbor (NN) search through query while the user is moving. NN search in such a dynamic environment has been performed through the repeated applicaton of the NN method to the search segment, but this increases search cost because of unnecessary redundant calculation. We propose slabbed continuous nearest neighbor (Slabbed_CNN) search, which is a new method that searches CNN in the search segment while moving, Slabbed_CNN reduces calculation costs and provides faster services than existing CNN by reducing the search area and calculation cost of the existing CNN method through reducing the search segment using slabs.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.27 no.1
• /
• pp.22-28
• /
• 2017

The Convolutional Neural Network(CNN) can be divided into two stages: feature extraction and classification. The hyperparameters such as kernel size, number of channels, and stride in the feature extraction step affect the overall performance of CNN as well as determining the structure of CNN. In this paper, we propose a method to optimize the hyperparameter in CNN feature extraction stage using Parameter-Setting-Free Harmony Search (PSF-HS) algorithm. After setting the overall structure of CNN, hyperparameter was set as a variable and the hyperparameter was optimized by applying PSF-HS algorithm. The simulation was conducted using MATLAB, and CNN learned and tested using mnist data. We update the parameters for a total of 500 times, and it is confirmed that the structure with the highest accuracy among the CNN structures obtained by the proposed method classifies the mnist data with an accuracy of 99.28%.

• Korean Journal of Remote Sensing
• /
• v.35 no.6_1
• /
• pp.959-971
• /
• 2019

CNN (Convolutional Neural Network) is one representative deep learning algorithm, which can extract high-level spatial and spectral features, and has been applied for hyperspectral image classification. However, one significant drawback behind the application of CNNs in hyperspectral images is the high dimensionality of the data, which increases the training time and processing complexity. To address this problem, several CNN based hyperspectral image classification studies have exploited PCA (Principal Component Analysis) for dimensionality reduction. One limitation to this is that the spectral information of the original image can be lost through PCA. Although it is clear that the use of PCA affects the accuracy and the CNN training time, the impact of PCA for CNN based hyperspectral image classification has been understudied. The purpose of this study is to analyze the quantitative effect of PCA in CNN for hyperspectral image classification. The hyperspectral images were first transformed through PCA and applied into the CNN model by varying the size of the reduced dimensionality. In addition, 2D-CNN and 3D-CNN frameworks were applied to analyze the sensitivity of the PCA with respect to the convolution kernel in the model. Experimental results were evaluated based on classification accuracy, learning time, variance ratio, and training process. The size of the reduced dimensionality was the most efficient when the explained variance ratio recorded 99.7%~99.8%. Since the 3D kernel had higher classification accuracy in the original-CNN than the PCA-CNN in comparison to the 2D-CNN, the results revealed that the dimensionality reduction was relatively less effective in 3D kernel.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.12
• /
• pp.587-594
• /
• 2021

As the diagnosis using encephalography(EEG) has been expanded, various studies have been actively performed for classifying EEG automatically. This paper proposes a CNN model that can effectively classify EEG signals acquired from healthy persons and patients with epilepsy. We segment the EEG signals into sub-signals with smaller dimension to augment the EEG data that is necessary to train the CNN model. Then the sub-signals are segmented again with overlap and they are used for training the CNN model. We also propose ensemble strategy in order to improve the classification accuracy. Experimental result using public Bonn dataset shows that the CNN can detect the epileptic seizure with the accuracy above 99.0%. It also shows that the ensemble method improves the accuracy of 3-class and 5-class EEG classification.

• The Journal of the Korea Contents Association
• /
• v.20 no.3
• /
• pp.267-274
• /
• 2020

Click-through rate (CTR) prediction is an estimate of the probability that a user will click on a given item and plays an important role in determining strategies for maximizing online ad revenue. Recently, research has been performed to utilize CNN for CTR prediction. Since the CTR data does not have a meaningful order in terms of correlation, the CTR data may be arranged in any order. However, because CNN only learns local information limited by filter size, data arrays can have a significant impact on performance. In this paper, we propose a multi-array CNN model that generates a data array set that can extract all local feature information that CNN can collect, and learns features through individual CNN modules. Experimental results for large data sets show that the proposed model achieves a 22.6% synergy with RI in AUC compared to the existing CNN, and the proposed array generation method achieves 3.87% performance improvement over the random generation method.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.79-83
• /
• 2018

본 논문은 한국어 문장의 형태소, 음절, 자소를 동시에 각자 다른 합성곱층을 통과시켜 문장의 감성을 분류하는 Multi-channel CNN을 제안한다. 오타를 포함하는 구어체 문장들의 경우에 형태소 기반 CNN으로 추출 할 수 없는 특징들을 음절이나 자소에서 추출 할 수 있다. 한국어 감성분석에 형태소 기반 CNN이 많이 쓰이지만, 본 논문의 Multi-channel CNN 모델은 형태소, 음절, 자소를 동시에 고려하여 더 정확하게 문장의 감성을 분류한다. 본 논문이 제안하는 모델이 형태소 기반 CNN보다 야구 댓글 데이터에서는 약 4.8%, 영화 리뷰 데이터에서는 약 1.3% 더 정확하게 문장의 감성을 분류하였다.