CNN-based Deep Learning Research Articles

Exploring Deep Ensemble Model for Insect and Pest Detection from Images

Pests and insects are the main hazards to crops, and they have a significant negative impact on both human health and wealth. Identification and classification of different types of crop insects accurately is an important task to avoid these losses. A manual examination is a quite sluggish and less efficient way of accomplishing this task while computer vision technology can significantly assist in this issue. A CNN-based deep learning model can classify insects and pests efficiently. In this paper, an ensemble-based model has been proposed using transfer learning. The experimental setup comprises pre-trained models like VGG16, VGG19, and ResNetv50 with a voting classifier ensemble technique. These pre-trained models are used to train the training dataset in a parallel pipeline model which is united with Ensemble Voting Classifier to generate the final prediction over the input sample. The benchmark IP102 dataset, having more than 75000 samples over 102 classes, is used to train and evaluate the model's performance. Results show that the proposed ensemble model achieves an accuracy of 82.5%, which is significant improvement over the recent state-of-the-art models proposed over the same selected dataset, hence strongly supporting the efficacy of the model.

A fine-grained deep learning model using embedded-CNN with BiLSTM for exploiting product sentiments

As technology advances, Facebook, Twitter, and microblogging sites have become the most effective platforms for communication and information exchange. Through these forums, people can share their views and experiences. These platforms enable discussion about a certain product that can be a valuable resource used to inform any decision-making process. For such studies, the majority of advanced-level researchers employed deep learning and machine learning models in conjunction with natural language processing (NLP). In recent years, the use of pre-trained models, such as Glove and BERT, in aspect-based sentiment analysis (ABSA) has increased. In ABSA, the auxiliary information is required to distinguish each aspect of this fine-grained task. However, BERT’s input format is restricted to a collection of words that cannot include more context knowledge. To address this problem, a BiLSTM and embedded CNN-based deep learning model has been presented for sentiment analysis at the aspect level. Initially, datasets were compiled from several sources. Then, an auxiliary feature was extracted using standard NLP. The auxiliary features were further refined and transformed into feature vectors based on the proposed embedded CNN model. Finally, a BiLSTM-based classification has been performed for sentiment classification. The experimental evaluation demonstrated that the performance of the suggested technique achieves on the SemEval dataset in terms of F1 score and accuracy by 81.7 and 83.3 percentage points, respectively, and on the product review dataset by 80.8 and 83.1 percentage points, respectively.

CNN-based object detection and growth estimation of plum fruit (Prunus mume) using RGB and depth imaging techniques

Modern people who value healthy eating habits have shown increasing interest in plum (Prunus mume) fruits, primarily owing to their nutritiousness and proven efficacy. As consumption increases, it becomes important to monitor work to prevent Prunus mume fruits from falling out. Moreover, determining the growth status of Prunus mume is also crucial and is attracting increasing attention. In this study, convolutional neural network (CNN)-based deep learning object detection was developed using RGBD images collected from Prunus mume farms. These RGBD images consider various environments, including the depth information of objects in the outdoor field. A faster region-based convolutional neural network (R-CNN), EfficientDet, Retinanet, and Single Shot Multibox Detector (SSD) were applied for detection, and the performance of all models was estimated by comparing their respective computing speeds and average precisions (APs). The test results show that the EfficientDet model is the most accurate, and SSD MobileNet is the fastest among the four models. In addition, the algorithm was developed to acquire the growth status of P. mume fruits by applying the coordinates and score values of bounding boxes to the depth map. Compared to the diameters of the artificial Prunus mume fruits used as the experimental group, the calculated diameters were very similar to those of the artificial objects. Collectively, the results demonstrate that the CNN-based deep learning Prunus mume detection and growth estimation method can be applied to real farmlands.

Emotion recognition while applying cosmetic cream using deep learning from EEG data; cross-subject analysis.

We report a deep learning-based emotion recognition method using EEG data collected while applying cosmetic creams. Four creams with different textures were randomly applied, and they were divided into two classes, "like (positive)" and "dislike (negative)", according to the preference score given by the subject. We extracted frequency features using well-known frequency bands, i.e., alpha, beta and low and high gamma bands, and then we created a matrix including frequency and spatial information of the EEG data. We developed seven CNN-based models: (1) inception-like CNN with four-band merged input, (2) stacked CNN with four-band merged input, (3) stacked CNN with four-band parallel input, and stacked CNN with single-band input of (4) alpha, (5) beta, (6) low gamma, and (7) high gamma. The models were evaluated by the Leave-One-Subject-Out Cross-Validation method. In like/dislike two-class classification, the average accuracies of all subjects were 73.2%, 75.4%, 73.9%, 68.8%, 68.0%, 70.7%, and 69.7%, respectively. We found that the classification performance is higher when using multi-band features than when using single-band feature. This is the first study to apply a CNN-based deep learning method based on EEG data to evaluate preference for cosmetic creams.

Classification of Cancer Images with CNN-Based Deep Learning Approach

İnsanların güneşin zararlı ışınlarının sebebiyet verdiği etkiler sonucunda halk arasında melanoma (mel), dermafibroma (df), ve vascular (vasc), bening keratosis (bkl), melanocytic nevi (nv), basal cell carcinoma (bcc), actinic keratosis (akiec) olarak bilinen türler sıklıkla görülmektedir. Biyomedikal cihazlar kullanılarak elde edilen cilt görüntüleri üzerinde yapılan derin öğrenme analizleri ile girdi olarak verilen cilt görüntüsünün kanser ise hangi cilt kanseri olduğu belirlenebilmektedir. Bu maksatla cilt alanında çalışan sağlık uzmanlarının iş yoğunluğunu düşürecek, hızlı tanı ve sınıflandırma yapabilecek CNN tabanlı bir derin öğrenme modeli önerilmiştir. Swish ve ReLU aktivasyon fonksiyonunun avantajlarından faydalanılarak geliştirilen CNN modeli akiec, bcc, bkl, df, nv, vasc, mel cilt kanseri sınıf türlerinde 0.99%, 0.99%, 0.96%, 0.99%, 0.92%, 0.99%, 0.95% F1 score değerlerini elde etmiştir

딥러닝 기반의 반도체 패키지 다이면 스크래치 검출 방법

The purpose of this paper is to evaluate the detection capability of deep learning-based surface scratches in semiconductor package dies and to determine whether deep learning can be employed as an objective inspection tool. To achieve this aim, detection capabilities are compared and evaluated based on a conventional image processing-based appearance inspection algorithm for examining equipment operating in the semiconductor package inspection process. In this study, a ResNet, which is the widely used CNN-based deep learning model, is used. The dataset was generated by collecting images of semiconductor packages being inspected in the semiconductor package appearance inspection process, which were cropped to focus on the parts necessary for the study. Three experiments were conducted for comparison. In the first experiment, the results of each detection using the Hough transform and ResNet-50 were compared. In the second experiment, the comparison was conducted in the same environment as in the first experiment by adding over-examination images. In the third experiment, the Hough transform algorithm and ResNet-50 were further compared by learning the over-examination images. Based on the results, an objective evaluation was conducted to determine whether deep learning could be applied to image processing-based machine inspection facilities.

Utility of the deep learning technique for the diagnosis of orbital invasion on CT in patients with a nasal or sinonasal tumor

BackgroundIn nasal or sinonasal tumors, orbital invasion beyond periorbita by the tumor is one of the important criteria in the selection of the surgical procedure. We investigated the usefulness of the convolutional neural network (CNN)-based deep learning technique for the diagnosis of orbital invasion, using computed tomography (CT) images.MethodsA total of 168 lesions with malignant nasal or sinonasal tumors were divided into a training dataset (n = 119) and a test dataset (n = 49). The final diagnosis (invasion-positive or -negative) was determined by experienced radiologists who carefully reviewed all of the CT images. In a CNN-based deep learning analysis, a slice of the square target region that included the orbital bone wall was extracted and fed into a deep-learning training session to create a diagnostic model using transfer learning with the Visual Geometry Group 16 (VGG16) model. The test dataset was subsequently tested in CNN-based diagnostic models and by two other radiologists who were not specialized in head and neck radiology. At approx. 2 months after the first reading session, two radiologists again reviewed all of the images in the test dataset, referring to the diagnoses provided by the trained CNN-based diagnostic model.ResultsThe diagnostic accuracy was 0.92 by the CNN-based diagnostic models, whereas the diagnostic accuracies by the two radiologists at the first reading session were 0.49 and 0.45, respectively. In the second reading session by two radiologists (diagnosing with the assistance by the CNN-based diagnostic model), marked elevations of the diagnostic accuracy were observed (0.94 and 1.00, respectively).ConclusionThe CNN-based deep learning technique can be a useful support tool in assessing the presence of orbital invasion on CT images, especially for non-specialized radiologists.

Automated detection of premalignant oral lesions on whole slide images using convolutional neural networks

IntroductionOral epithelial dysplasia (OED) is a precursor lesion to oral squamous cell carcinoma, a disease with a reported overall survival rate of 56 percent across all stages. Accurate detection of OED is critical as progression to oral cancer can be impeded with complete excision of premalignant lesions. However, previous research has demonstrated that the task of grading of OED, even when performed by highly trained experts, is subject to high rates of reader variability and misdiagnosis. Thus, our study aims to develop a convolutional neural network (CNN) model that can identify regions suspicious for OED whole-slide pathology images. MethodsDuring model development, we optimized key training hyperparameters including loss function on 112 pathologist annotated cases between the training and validation sets. Then, we compared OED segmentation and classification metrics between two well-established CNN architectures for medical imaging, DeepLabv3+ and UNet++. To further assess generalizability, we assessed case-level performance of a held-out test set of 44 whole-slide images. ResultsDeepLabv3+ outperformed UNet++ in overall accuracy, precision, and segmentation metrics in a 4-fold cross validation study. When applied to the held-out test set, our best performing DeepLabv3+ model achieved an overall accuracy and F1-Score of 93.3 percent and 90.9 percent, respectively. ConclusionThe present study trained and implemented a CNN-based deep learning model for identification and segmentation of oral epithelial dysplasia (OED) with reasonable success. Computer assisted detection was shown to be feasible in detecting premalignant/precancerous oral lesions, laying groundwork for eventual clinical implementation.

Comparative performance of four CNN-based deep learning variants in detecting Hispa pest, two fungal diseases, and NPK deficiency symptoms of rice (Oryza sativa)

Crop production can be significantly increased if stresses are detected at the earliest possible time to facilitate the implementation of necessary mitigation measures. This present study aims to evaluate the comparative performance of the Convolutional Neural Network (CNN) and four pre-trained deep CNN models, for automatic and rapid detections of Hispa, brown spot, leaf blast, and NPK deficiency symptoms from public and real field images. Deep learning models were trained using different public and field datasets combinations. Best accuracy was achieved for mixed public and field datasets rather than solely field or solely public datasets, with VGG19 models achieving 91.8% accuracy. Relatively simple structured CNN was found to predict phosphate deficient leaves with better performance (96% accuracy) than the other four advanced models, while VGG16 performed better for leaf blast and N deficiency identification. Likewise, ResNet50 could be recommended among the five models for the Potassium deficient leaf identification, while for the Hispa pest, VGG19 outperforms the InceptionV3, ResNet50, and VGG16 models, and was slightly better than the basic CNN. The implication of the study is enormous considering practical application as it deals with the complex dataset of pest, disease and nutrient deficiency. This non-invasive way of detecting different stresses of rice could help farmers in practicing precision agriculture. The findings from this study could be used to develop a user-friendly interface for the rapid and inexpensive detection of diseases and nutrition status by farmers, in a non-destructive way. Similar to rice, various biotic and abiotic stress symptoms in other economic crops could also be captured and identified by deep learning algorithms.

Recovery of impenetrable rough surface profiles via CNN-based deep learning architecture

ABSTRACT In this paper, a convolutional neural network (CNN)-based deep learning (DL) architecture for the solution of an electromagnetic inverse problem related to imaging of the shape of the perfectly electric conducting (PEC) rough surfaces is addressed. The rough surface is illuminated by a plane wave and scattered field data is obtained synthetically through the numerical solution of surface integral equations. An effective CNN-DL architecture is implemented through the modelling of the rough surface variation in terms of convenient spline type base functions. The algorithm is numerically tested with various scenarios including amplitude only data and shown that it is very effective and useful.

Convolutional Neural Network Techniques for Brain Tumor Classification (from 2015 to 2022): Review, Challenges, and Future Perspectives.

Convolutional neural networks (CNNs) constitute a widely used deep learning approach that has frequently been applied to the problem of brain tumor diagnosis. Such techniques still face some critical challenges in moving towards clinic application. The main objective of this work is to present a comprehensive review of studies using CNN architectures to classify brain tumors using MR images with the aim of identifying useful strategies for and possible impediments in the development of this technology. Relevant articles were identified using a predefined, systematic procedure. For each article, data were extracted regarding training data, target problems, the network architecture, validation methods, and the reported quantitative performance criteria. The clinical relevance of the studies was then evaluated to identify limitations by considering the merits of convolutional neural networks and the remaining challenges that need to be solved to promote the clinical application and development of CNN algorithms. Finally, possible directions for future research are discussed for researchers in the biomedical and machine learning communities. A total of 83 studies were identified and reviewed. They differed in terms of the precise classification problem targeted and the strategies used to construct and train the chosen CNN. Consequently, the reported performance varied widely, with accuracies of 91.63–100% in differentiating meningiomas, gliomas, and pituitary tumors (26 articles) and of 60.0–99.46% in distinguishing low-grade from high-grade gliomas (13 articles). The review provides a survey of the state of the art in CNN-based deep learning methods for brain tumor classification. Many networks demonstrated good performance, and it is not evident that any specific methodological choice greatly outperforms the alternatives, especially given the inconsistencies in the reporting of validation methods, performance metrics, and training data encountered. Few studies have focused on clinical usability.

Development of a convolutional neural network joint detector for non-orthogonal multiple access uplink receivers

We present a novel approach to signal detection for Non-Orthogonal Multiple Access (NOMA) uplink receivers using Convolutional Neural Networks (CNNs) in a single-shot fashion. The defacto NOMA detection method is the so-called Successive Interference Cancellation which requires precise channel estimation and accurate successive detection of the user equipment with the higher powers. It is proposed converting incoming packets into 2D image-like streams. These images are fed to a CNN-based deep learning network commonly used in the image processing literature for image classification. The classification label for each packet converted to an image is the transmitted symbols by all user equipment joined together. CNN network is trained using uniformly distributed samples of incoming packets at different signals to noise ratios. Furthermore, let’s performed hyperparameter optimization using the exhaustive search method. Our approach is tested using a modeled system of two user equipment systems in a 64-subcarrier Orthogonal Frequency Division Multiplexing (OFDM) and Rayleigh channel. It is found that a three-layer CNN with 32 filters of size 7×7 has registered the highest training and testing accuracy of about 81. In addition, our result showed significant improvement in Symbol Error Rate (SER) vs. Signal to Noise Ratio (SNR) compared to other state-of-the-art approaches such as least square, minimum mean square error, and maximum likelihood under various channel conditions. When the channel length is fixed at 20, our approach is at least one significant Figure better than the maximum likelihood method at (SNR) of 2 dB. Finally, the channel length to 12 is varied and it is registered about the same performance. Hence, our approach is more robust to joint detection in NOMA receivers, particularly in low signal-to-noise environments

Depth Estimation for Integral Imaging Microscopy Using a 3D–2D CNN with a Weighted Median Filter

This study proposes a robust depth map framework based on a convolutional neural network (CNN) to calculate disparities using multi-direction epipolar plane images (EPIs). A combination of three-dimensional (3D) and two-dimensional (2D) CNN-based deep learning networks is used to extract the features from each input stream separately. The 3D convolutional blocks are adapted according to the disparity of different directions of epipolar images, and 2D-CNNs are employed to minimize data loss. Finally, the multi-stream networks are merged to restore the depth information. A fully convolutional approach is scalable, which can handle any size of input and is less prone to overfitting. However, there is some noise in the direction of the edge. A weighted median filtering (WMF) is used to acquire the boundary information and improve the accuracy of the results to overcome this issue. Experimental results indicate that the suggested deep learning network architecture outperforms other architectures in terms of depth estimation accuracy.

CNN-Based Deep Learning Technique for the Brain Tumor Identification and Classification in MRI Images

A brain tumor is an abnormal development of cells in the brain that are either benign or malignant. Magnetic resonance imaging (MRI) is used to identify tumors. Manual evaluation of brain tumors from MRI images by a radiologist is a challenging task. Hence, this paper proposes VGG-19 Convolutional Neural Networks (CNN)-based deep learning model for the classification of brain tumors. Initially, in the proposed model, contrast stretching technique is employed for noises removal. Next, a deep neural network is employed for rich feature extract. Further, these learning features are combined with classifier models of CNN for training and validation. performance analysis of the proposed methodology and experiments have been carried out using publicly available MRI images in Figshare dataset of 3064 slices from 233 subjects. The proposed model has achieved 99.83% accuracy. Moreover, the proposed model obtained precision 96.32%, 98.26%, and 98.56%, recall of 97.82%, 98.62%, 98.87%, and specificity of 98.72%, 99.51%, and 99.43% for the Glioma, Meningioma, and Pituitary tumors respectively.

Semantic Segmentation of Extraocular Muscles on Computed Tomography Images Using Convolutional Neural Networks.

Computed tomography (CT) imaging of the orbit with measurement of extraocular muscle size can be useful for diagnosing and monitoring conditions that affect extraocular muscles. However, the manual measurement of extraocular muscle size can be time-consuming and tedious. The purpose of this study is to evaluate the effectiveness of deep learning algorithms in segmenting extraocular muscles and measuring muscle sizes from CT images. Consecutive CT scans of orbits from 210 patients between 1 January 2010 and 31 December 2019 were used. Extraocular muscles were manually annotated in the studies, which were then used to train the deep learning algorithms. The proposed U-net algorithm can segment extraocular muscles on coronal slices of 32 test samples with an average dice score of 0.92. The thickness and area measurements from predicted segmentations had a mean absolute error (MAE) of 0.35 mm and 3.87 mm2, respectively, with a corresponding mean absolute percentage error (MAPE) of 7 and 9%, respectively. On qualitative analysis of 32 test samples, 30 predicted segmentations from the U-net algorithm were accepted while 2 were rejected. Based on the results from quantitative and qualitative evaluation, this study demonstrates that CNN-based deep learning algorithms are effective at segmenting extraocular muscles and measuring muscles sizes.

Deep Learning Image Classification for Fashion Design

Fashion has always been an essential feature in our daily routine. It also plays a significant role in everyone’s lives. In this research, convolutional neural networks (CNN) were used to train images of different fashion styles, which were attempted to be predicted with a high success rate. Deep learning has been widely applied in a variety of fields recently. A CNN is a deep neural network that delivers the most accurate answers when tackling real-world situations. Apparel manufacturers have employed CNN to tackle various difficulties on their e-commerce sites, including clothing recognition, search, and suggestion. A set of photos from the Fashion-MNIST dataset is used to train a series of CNN-based deep learning architectures to distinguish between photographs. CNN design, batch normalization, and residual skip connections reduce the time it takes to learn. The CNN model’s findings are evaluated using the Fashion-MNIST datasets. In this paper, classification is done with a convolutional layer, filter size, and ultimately connected layers. Experiments are run with different activation functions, optimizers, learning rates, dropout rates, and batch sizes. The results showed that the choice of activation function, optimizer, and dropout rate impacts the correctness of the results.

Classifier identification using deep learning and machine learning algorithms for the detection of valvular heart diseases

Heart-related disorders are rapidly growing throughout the world. Artificial Intelligence with computational methods plays a significant role in early detection and diagnosis. This study has been devoted to finding the best classifiers for different valvular heart problems using popular CNN-based deep learning models and machine learning algorithms written in Python 3.8. In this research, the CNN-based Xception network model for the first time has been proposed for valvular heart sound analysis, which achieved an accuracy of 99.45% on the test dataset with a sensitivity of 98.5% and specificity of 98.7%. Compared with other deep learning models like LeNet-5, AlexNet, VGG16, VGG19, DenseNet121, Inception Net, and Residual Net, it is observed that accuracy for predicting the prediction of valvular heart disease is the highest, and testing time is the lowest in the proposed modified Xception network model. The features are Root Mean Square, Energy, Power, Zero Crossing Rate, Total Harmonic Distortion, Skewness, and Kurtosis in the time domain. The analysis has been made on heart sounds of normal and diseased patients available from the standard heart sound data repository. Finally, all the evaluated results were compared, and found SVM and Random Forest algorithms are the most effective among machine learning methods. The proposed modified CNN-based Xception model works the best among all deep learning methods.

A CNN-Based Deep Learning Approach for SSVEP Detection Targeting Binaural Ear-EEG

This paper discusses a machine learning approach for detecting SSVEP at both ears with minimal channels. SSVEP is a robust EEG signal suitable for many BCI applications. It is strong at the visual cortex around the occipital area, but the SNR gets worse when detected from other areas of the head. To make use of SSVEP measured around the ears following the ear-EEG concept, especially for practical binaural implementation, we propose a CNN structure coupled with regressed softmax outputs to improve accuracy. Evaluating on a public dataset, we studied classification performance for both subject-dependent and subject-independent trainings. It was found that with the proposed structure using a group training approach, a 69.21% accuracy was achievable. An ITR of 6.42 bit/min given 63.49 % accuracy was recorded while only monitoring data from T7 and T8. This represents a 12.47% improvement from a single ear implementation and illustrates potential of the approach to enhance performance for practical implementation of wearable EEG.

Application of Convolutional Neural Network-Based Detection Methods in Fresh Fruit Production: A Comprehensive Review.

As one of the representative algorithms of deep learning, a convolutional neural network (CNN) with the advantage of local perception and parameter sharing has been rapidly developed. CNN-based detection technology has been widely used in computer vision, natural language processing, and other fields. Fresh fruit production is an important socioeconomic activity, where CNN-based deep learning detection technology has been successfully applied to its important links. To the best of our knowledge, this review is the first on the whole production process of fresh fruit. We first introduced the network architecture and implementation principle of CNN and described the training process of a CNN-based deep learning model in detail. A large number of articles were investigated, which have made breakthroughs in response to challenges using CNN-based deep learning detection technology in important links of fresh fruit production including fruit flower detection, fruit detection, fruit harvesting, and fruit grading. Object detection based on CNN deep learning was elaborated from data acquisition to model training, and different detection methods based on CNN deep learning were compared in each link of the fresh fruit production. The investigation results of this review show that improved CNN deep learning models can give full play to detection potential by combining with the characteristics of each link of fruit production. The investigation results also imply that CNN-based detection may penetrate the challenges created by environmental issues, new area exploration, and multiple task execution of fresh fruit production in the future.

Leukemia classification using the deep learning method of CNN.

Processing Low-Intensity Medical Images (LI-MI) is difficult as outcomes are varied when it comes to manual examination, which is also a time-consuming process. To improve the quality of low-intensity images and identify the leukemia classification by utilizing CNN-based Deep Learning (DCNN) strategy. The strategies employed for the recognition of leukemia classifications in the advised strategy are DCNN (ResNet-34 & DenseNet-121). The histogram equalization-based adaptive gamma correction followed by guided filtering applies to study the improvement in intensity and preserve the essential details of the image. The DCNN is used as a feature extractor to help classify leukemia types. Two datasets of ASH with 520 images and ALL-IDB with 559 images are used in this study. In 1,079 images, 779 are positive cases depicting leukemia and 300 images are negative (normal) cases. Thus, to validate performance of this DCNN strategy, ASH and ALL-IDB datasets are promoted in the investigation process to classify between positive and negative images. The DCNN classifier yieldes the overall classification accuracy of 99.2% and 98.4%, respectively. In addition, the achieved classification specificity, sensitivity, and precision are 99.3%, 98.7%, 98.4%, and 98.9%, 98.4%,98.6% applying to two datasets, respectively, which are higher than the performance using other machine learning classifiers including support vector machine, decision tree, naive bayes, random forest and VGG-16. Ths study demonstrates that the proposed DCNN enables to improve low-intensity images and accuracry of leukemia classification, which is superior to many of other machine leaning classifiers used in this research field.