Deep Learning-based Image Recognition Research Articles

Computed tomography-based prediction of pancreatitis following biliary metal stent placement with the convolutional neural network.

Background and study aims Pancreatitis is a potentially lethal adverse event of endoscopic transpapillary placement of a self-expandable metal stent (SEMS) for malignant biliary obstruction (MBO). Deep learning-based image recognition has not been investigated in predicting pancreatitis in this setting. Patients and methods We included 70 patients who underwent endoscopic placement of a SEMS for nonresectable distal MBO. We constructed a convolutional neural network (CNN) model for pancreatitis prediction using a series of pre-procedure computed tomography images covering the whole pancreas (≥ 120,960 augmented images in total). We examined the additional effects of the CNN-based probabilities on the following machine learning models based on clinical parameters: logistic regression, support vector machine with a linear or RBF kernel, random forest classifier, and gradient boosting classifier. Model performance was assessed based on the area under the curve (AUC) in the receiver operating characteristic analysis, positive predictive value (PPV), accuracy, and specificity. Results The CNN model was associated with moderate levels of performance metrics: AUC, 0.67; PPV, 0.45; accuracy, 0.66; and specificity, 0.63. When added to the machine learning models, the CNN-based probabilities increased the performance metrics. The logistic regression model with the CNN-based probabilities had an AUC of 0.74, PPV of 0.85, accuracy of 0.83, and specificity of 0.96, compared with 0.72, 0.78, 0.77, and 0.96, respectively, without the probabilities. Conclusions The CNN-based model may increase predictability for pancreatitis following endoscopic placement of a biliary SEMS. Our findings support the potential of deep learning technology to improve prognostic models in pancreatobiliary therapeutic endoscopy.

Efficient Algorithmic Models for Improved Adversarial Attacks and Defenses in Deep Learning Based Image Recognition Models

Abstract: This paper delves into the critical concern of ensuring the safety as well as the reliability of models for deep learning amidst the escalating landscape of adversarial attacks. Techniques like FGSM, DeepFool and PGD pose substantial threats by manipulating input data, leading to erroneous outcomes within machine learning systems. Addressing this challenge head-on, our study introduces an innovative model explicitly engineered to counteract these adversarial threats. Our model specifically focuses on combating the notorious attack algorithms of FGSM, DeepFool and PGD by implementing robust defense mechanisms such as Adversarial Training and GANs. Through meticulous evaluations spanning diverse datasets, including CIFAR and MNIST, our model's efficacy in defending against these sophisticated attacks was rigorously assessed. The empirical results underscore the resilience of our model, showcasing its effectiveness in fortifying deep learning frameworks against hostile intrusions across varied datasets. Our research contributes crucial insights and formidable defense mechanisms, augmenting the security, trust, and reliability of these systems, even amidst the complex challenges posed by adversarial manipulations.

Application and evaluation of deep learning based image recognition techniques in agriculture

Agriculture, as an important industry in society, is facing problems such as an aging population and rural labor exodus, which leads to rising labor costs and uncertainty in agricultural production. Deep learning techniques are considered as a key tool to solve this problem. In this paper, three popular deep learning algorithms, namely, Region-based Convolutional Neural Network, You Only Look Once, and Single Shot MultiBox Detector, are introduced and their working principles are described in detail, while the advantages and disadvantages of these algorithms are briefly analyzed. Additionally, this paper specifically analyzes the application of these three algorithms in three agricultural scenarios, such as timber species recognition, fruit picking, and pest identification. The results show that although the three algorithms are slightly different in terms of accuracy and detection speed, they all demonstrate the potential for a wide range of applications in the agricultural field. Therefore, deep learning technology is of great significance in solving the problem of rural labor shortage, especially when combined with advanced equipment, which is expected to significantly improve the efficiency of identification, monitoring, and harvesting in agriculture and promote the development of automated agriculture.

Application of image recognition technology based on deep learning in plants disease detection and diagnosis

With increasing demographic trends, the world will likely confront a looming food crisis in the coming decades. This paper points out the world's food crisis brought by the increasing population and the agricultural sector's challenges. Among them, the increase in plant diseases is considered one of the leading causes of agricultural inefficiency and food production problems. For example, in many countries, cash crops such as jute, pomegranate, and tomato have been severely affected by plant diseases, resulting in reduced yields and, thus, increased economic losses. Traditional disease detection methods, such as manual visual inspection and laboratory analysis, are inaccurate and time-consuming, requiring a lot of manpower and capital investment. In recent years, researchers have begun to apply deep learning-based image recognition technology to detect and diagnose plant diseases to improve accuracy and efficiency. This paper will lay the foundation for further research by summarizing the existing research and analyzing the advantages and applicability of different application methods.

ResNet incorporating the fusion data of RGB & hyperspectral images improves classification accuracy of vegetable soybean freshness

The freshness of vegetable soybean (VS) is an important indicator for quality evaluation. Currently, deep learning-based image recognition technology provides a fast, efficient, and low-cost method for analyzing the freshness of food. The RGB (red, green, and blue) image recognition technology is widely used in the study of food appearance evaluation. In addition, the hyperspectral image has outstanding performance in predicting the nutrient content of samples. However, there are few reports on the research of classification models based on the fusion data of these two sources of images. We collected RGB and hyperspectral images at four different storage times of VS. The ENVI software was adopted to extract the hyperspectral information, and the RGB images were reconstructed based on the downsampling technology. Then, the one-dimensional hyperspectral data was transformed into a two-dimensional space, which allows it to be overlaid and concatenated with the RGB image data in the channel direction, thereby generating fused data. Compared with four commonly used machine learning models, the deep learning model ResNet18 has higher classification accuracy and computational efficiency. Based on the above results, a novel classification model named ResNet-R &H, which is based on the residual networks (ResNet) structure and incorporates the fusion data of RGB and hyperspectral images, was proposed. The ResNet-R &H can achieve a testing accuracy of 97.6%, which demonstrates a significant enhancement of 4.0% and 7.2% compared to the distinct utilization of hyperspectral data and RGB data, respectively. Overall, this research is significant in providing a unique, efficient, and more accurate classification approach in evaluating the freshness of vegetable soybean. The method proposed in this study can provide a theoretical reference for classifying the freshness of fruits and vegetables to improve classification accuracy and reduce human error and variability.

Deep Learning Based Image Recognition Technology for Civil Engineering Applications

Abstract In this paper, we use Caffe framework to implement the improved Faster R-CNN recognition technique for building images in civil engineering under Linux system and add feature pyramid network and regional feature aggregation into the ResNet-50 network and ResNet-101 network, respectively, to strengthen the training effect, and establish ResNet-101+FPN+ROI Align image recognition technique. Simulated crack experiments and concrete surface quality defect detection experiments confirm that the ResNet-101 FPN ROI Align method is accurate and detects defects at a high rate. The method established in this paper has a minimum error of only 0.4% in the simulated crack experiment, and the detection rate is much higher than that of other detection methods when detecting quality defects on the concrete surface, and the accuracy can reach up to 94% at the same time. In civil engineering, the image recognition technology established in this paper has practical significance and high application value, as demonstrated by the experiment.

Deep Learning Image Recognition-Assisted Atomic Force Microscopy for Single-Cell Efficient Mechanics in Co-culture Environments.

Atomic force microscopy (AFM)-based force spectroscopy assay has become an important method for characterizing the mechanical properties of single living cells under aqueous conditions, but a disadvantage is its reliance on manual operation and experience as well as the resulting low throughput. Particularly, providing a capacity to accurately identify the type of the cell grown in co-culture environments without the need of fluorescent labeling will further facilitate the applications of AFM in life sciences. Here, we present a study of deep learning image recognition-assisted AFM, which not only enables fluorescence-independent recognition of the identity of single co-cultured cells but also allows efficient downstream AFM force measurements of the identified cells. With the use of the deep learning-based image recognition model, the viability and type of individual cells grown in co-culture environments were identified directly from the optical bright-field images, which were confirmed by the following cell growth and fluorescent labeling results. Based on the image recognition results, the positional relationship between the AFM probe and the targeted cell was automatically determined, allowing the precise movement of the AFM probe to the target cell to perform force measurements. The experimental results show that the presented method was applicable not only to the conventional (microsphere-modified) AFM probe used in AFM indentation assay for measuring the Young's modulus of single co-cultured cells but also to the single-cell probe used in AFM-based single-cell force spectroscopy (SCFS) assay for measuring the adhesion forces of single co-cultured cells. The study illustrates deep learning imaging recognition-assisted AFM as a promising approach for label-free and high-throughput detection of single-cell mechanics under co-culture conditions, which will facilitate unraveling the mechanical cues involved in cell-cell interactions in their native states at the single-cell level and will benefit the field of mechanobiology.

An AI Dietitian for Type 2 Diabetes Mellitus Management Based on Large Language and Image Recognition Models: Preclinical Concept Validation Study.

Nutritional management for patients with diabetes in China is a significant challenge due to the low supply of registered clinical dietitians. To address this, an artificial intelligence (AI)-based nutritionist program that uses advanced language and image recognition models was created. This program can identify ingredients from images of a patient's meal and offer nutritional guidance and dietary recommendations. The primary objective of this study is to evaluate the competence of the models that support this program. The potential of an AI nutritionist program for patients with type 2 diabetes mellitus (T2DM) was evaluated through a multistep process. First, a survey was conducted among patients with T2DM and endocrinologists to identify knowledge gaps in dietary practices. ChatGPT and GPT 4.0 were then tested through the Chinese Registered Dietitian Examination to assess their proficiency in providing evidence-based dietary advice. ChatGPT's responses to common questions about medical nutrition therapy were compared with expert responses by professional dietitians to evaluate its proficiency. The model's food recommendations were scrutinized for consistency with expert advice. A deep learning-based image recognition model was developed for food identification at the ingredient level, and its performance was compared with existing models. Finally, a user-friendly app was developed, integrating the capabilities of language and image recognition models to potentially improve care for patients with T2DM. Most patients (182/206, 88.4%) demanded more immediate and comprehensive nutritional management and education. Both ChatGPT and GPT 4.0 passed the Chinese Registered Dietitian examination. ChatGPT's food recommendations were mainly in line with best practices, except for certain foods like root vegetables and dry beans. Professional dietitians' reviews of ChatGPT's responses to common questions were largely positive, with 162 out of 168 providing favorable reviews. The multilabel image recognition model evaluation showed that the Dino V2 model achieved an average F1 score of 0.825, indicating high accuracy in recognizing ingredients. The model evaluations were promising. The AI-based nutritionist program is now ready for a supervised pilot study.

Sooty Mold Detection on Citrus Tree Canopy Using Deep Learning Algorithms.

Sooty mold is a common disease found in citrus plants and is characterized by black fungi growth on fruits, leaves, and branches. This mold reduces the plant's ability to carry out photosynthesis. In small leaves, it is very difficult to detect sooty mold at the early stages. Deep learning-based image recognition techniques have the potential to identify and diagnose pest damage and diseases such as sooty mold. Recent studies used advanced and expensive hyperspectral or multispectral cameras attached to UAVs to examine the canopy of the plants and mid-range cameras to capture close-up infected leaf images. To bridge the gap on capturing canopy level images using affordable camera sensors, this study used a low-cost home surveillance camera to monitor and detect sooty mold infection on citrus canopy combined with deep learning algorithms. To overcome the challenges posed by varying light conditions, the main reason for using specialized cameras, images were collected at night, utilizing the camera's built-in night vision feature. A total of 4200 sliced night-captured images were used for training, 200 for validation, and 100 for testing, employed on the YOLOv5m, YOLOv7, and CenterNet models for comparison. The results showed that YOLOv7 was the most accurate in detecting sooty molds at night, with 74.4% mAP compared to YOLOv5m (72%) and CenterNet (70.3%). The models were also tested using preprocessed (unsliced) night images and day-captured sliced and unsliced images. The testing on preprocessed (unsliced) night images demonstrated the same trend as the training results, with YOLOv7 performing best compared to YOLOv5m and CenterNet. In contrast, testing on the day-captured images had underwhelming outcomes for both sliced and unsliced images. In general, YOLOv7 performed best in detecting sooty mold infections at night on citrus canopy and showed promising potential in real-time orchard disease monitoring and detection. Moreover, this study demonstrated that utilizing a cost-effective surveillance camera and deep learning algorithms can accurately detect sooty molds at night, enabling growers to effectively monitor and identify occurrences of the disease at the canopy level.

Assessment of deep learning-based image analysis for disaster waste identification

Accurately predicting different types and quantities of wastes is crucial for proper waste management and sustainable growth and development. High levels of disaster waste (DW) can negatively impact emergency responses, public health, and recovery and rebuilding processes, making the effective management of DW highly important. Most existing waste identification methods cannot suitably identify DW since it is highly subjective in form. In this study, we examined the applicability of different deep learning-based image recognition techniques for DW recognition. A DW image dataset was collected and categorized into normal and hard cases. Then, experiments were conducted using these three network models: DeepLabV3+, ResNeSt-50, and ResNeSt-101. The experimental results indicated that the mean intersection over union (mIoU) of the DeepLabV3+ model was 0.802 for the normal case and that of the ResNeSt-101 model was 0.676 for the hard case, indicating favorable performances. Overall, the deep learning-based image analysis method is more robust than existing methods, particularly in terms of its ability to accurately identify waste in various forms, such as mixed or piled up waste.

Development of an intelligent field investigation system for Liriomyza using SeResNet-Liriomyza for accurate identification

Liriomyza (Diptera: Agromyzidae) is genus of insects found worldwide. Among the species in this genus, L. sativae, L. huidobrensis, L. trifolii, and L. bryoniae are the four most globally invasive, with L. chinensis being native to China. Real-time and accurate monitoring is a prerequisite for reducing the spread and crop damage caused by this pest. Owing to the small size of Liriomyza, specialized technicians using a microscope are needed to identify them. We developed an intelligent field investigation system to conveniently and accurately identify these five species. The system includes a portable image acquisition device based on a smartphone, a deep learning-based image recognition model for Liriomyza, and a mobile platform-based investigation application. To capture magnified images of Liriomyza, we designed a magnifying lens and a rotatable carrier assembled outside the smartphone camera. To address the problem of local blurring in images captured by smartphones, all images were sharpened to enhance their features. To accurately distinguish these five species from similar species, we developed a SeResNet-Liriomyza model, which embeds spatial feature pyramid networks (SFPN) in the SeResNet model to prevent the false positives caused by different sizes of input images and generate new feature maps favorable for classification by combining global and local image features. Compared with the excellent classification models, DenseNet-201, ResNeXt-50, MobileNet-v3, EfficientNet-B4, and ResNeSt-101, SeResNet-Liriomyza showed excellent recognition performance, with an average recognition accuracy of 99.88%. We designed and developed a mobile application (app) for the intelligent investigation of Liriomyza. The app includes functions such as intelligent investigation, intelligent recognition, a biological dictionary, and investigation result retrieval. The proposed intelligent field investigation system provides investigators with a fast, convenient, and accurate tool for investigating Liriomyza, a pest that can cause serious damage to agriculture in the field.

Researches advanced in image recognization based on deep learning

Image recognition is a fundamental problem in the field of computer vision community, which aims to understand the content of an image to predict the class of the image. Traditional image recognition methods rely heavily on the quality of handcrafted features, and the recognition accuracy and generalization ability cannot meet practical application requirements. Thanks to the speedy progress of deep learning theory and technology, convolutional neural networks can adaptively learn image semantic features through high-dimensional nonlinear changes, which sensibly rises the exactness of image recognition. At present, image recognition has been generally applied to many fields such as commodity circulation, smart retail, pest identification, medical image analysis and so on. In this paper, we introduce the latest research progress in deep learning-based image recognition. Specifically, following the temporal clues of technological development, we first introduce representative image recognition networks, including their design ideas, basic network structures, advantages and disadvantages, etc. Second, we quantitatively compare the performance of different image recognition networks. Finally, we summarize the existing problems in the sphere of image recognition research and discuss its possible future directions.

Researches Advanced in Image Recognition based on Deep Learning

In the discipline of computer vision, which tries to create a model to predict the category of objects present in a given image, image recognition has always been a research hotspot. Early image recognition models were mainly based on manual features, and their recognition accuracy and generalization ability often fluctuated greatly with changes in the scene, which could not meet the actual application requirements. Convolutional neural networks have progressed quickly, which has accelerated the development of deep learning-based image recognition. in this paper, we detail the development of image recognition technology. Specifically, we introduce the three approaches of convolutional neural network, recurrent neural network, and graph neural network as the traditional methods of image recognition. including their design ideas, key steps, advantages and disadvantages. We also compare the recognition accuracy of different methods on the ImageNe dataset to try to explore the application boundaries of different methods. Finally, we go over the difficulties in the area of image identification and project its future growth.

Rethinking Breast Cancer Diagnosis through Deep Learning Based Image Recognition.

This paper explored techniques for diagnosing breast cancer using deep learning based medical image recognition. X-ray (Mammography) images, ultrasound images, and histopathology images are used to improve the accuracy of the process by diagnosing breast cancer classification and by inferring their affected location. For this goal, the image recognition application strategies for the maximal diagnosis accuracy in each medical image data are investigated in terms of various image classification (VGGNet19, ResNet50, DenseNet121, EfficietNet v2), image segmentation (UNet, ResUNet++, DeepLab v3), and related loss functions (binary cross entropy, dice Loss, Tversky loss), and data augmentation. As a result of evaluations through the presented methods, when using filter-based data augmentation, ResNet50 showed the best performance in image classification, and UNet showed the best performance in both X-ray image and ultrasound image as image segmentation. When applying the proposed image recognition strategies for the maximal diagnosis accuracy in each medical image data, the accuracy can be improved by 33.3% in image segmentation in X-ray images, 29.9% in image segmentation in ultrasound images, and 22.8% in image classification in histopathology images.

Review of application of Deep Learning based Image recognition in Disease diagnosis

Review of application of Deep Learning based Image recognition in Disease diagnosis

Headlight Control Using Deep Learning-Based Vehicle Detection

Abstract: Vehicle lights are the best assurance of driving safety when it is dark outside. High beams are frequently turned on by drivers to narrow their field of vision, to make the oncoming car more visible, or even when they are following another vehicle. The lights themselves hinder drivers from checking their rear vision in the rearview mirror and result in traffic collisions. Additionally, failure to maintain a safe distance from oncoming traffic by vehicles frequently results in accidents. This research suggests a deep learning-based image recognition system for the headlamp control system as a solution to this issue. When a driver is operating a car equipped with this system, they can instantly identify the vehicle in front of them and estimate at night the safety distance of the car in front of them to reduce light damage and traffic accidents at safe distances, consider if there are cars in the forward and opposing lanes before deciding whether to turn on the high beam.

Deep Learning-Based Image Recognition of Agricultural Pests

Pests and diseases are an inevitable problem in agricultural production, causing substantial economic losses yearly. The application of convolutional neural networks to the intelligent recognition of crop pest images has become increasingly popular due to advances in deep learning methods and the rise of large-scale datasets. However, the diversity and complexity of pest samples, the size of sample images, and the number of examples all directly affect the performance of convolutional neural networks. Therefore, we designed a new target-detection framework based on Cascade RCNN (Regions with CNN features), aiming to solve the problems of large image size, many pest types, and small and unbalanced numbers of samples in pest sample datasets. Specifically, this study performed data enhancement on the original samples to solve the problem of a small and unbalanced number of examples in the dataset and developed a sliding window cropping method, which could increase the perceptual field to learn sample features more accurately and in more detail without changing the original image size. Secondly, combining the attention mechanism with the FPN (Feature Pyramid Networks) layer enabled the model to learn sample features that were more important for the current task from both channel and space aspects. Compared with the current popular target-detection frameworks, the average precision value of our model (mAP@0.5) was 84.16%, the value of (mAP@0.5:0.95) was 65.23%, the precision was 67.79%, and the F1 score was 82.34%. The experiments showed that our model solved the problem of convolutional neural networks being challenging to use because of the wide variety of pest types, the large size of sample images, and the difficulty of identifying tiny pests.

High-throughput investigation of Nb and Ta alloying effects on the microstructure and properties of a novel Ni-Co-based superalloy

Recent research has focused on the development of disk superalloys with better temperature capability and mechanical properties via a new alloying concept that uses high levels of Co and γ′ formers of Al, Ti, Ta and Nb to produce composite precipitation strengthening. In this work, alloying effects of Nb and Ta on phase stability, microstructure and property of a novel Ni-Co-based superalloy were systematically investigated using the high-throughput multi-component diffusion multiples (MCDMs) in combination with various high-throughput measurement tools. A deep learning-based image recognition method was developed to efficiently extract the γ′ morphological features, including area fraction, particle size, number density. The rapid characterizations of alloy compositions, microstructure and Vickers hardness (HV) enable us to establish the composition-dependent microstructural characteristics and mechanical property as well as their correlation with the phase stability, which can be directly applied for designing Ni-Co-based superalloys with superior microstructure stability and strength.

Advances in deep learning-based image recognition of product packaging

The information about the goods on the shelves needed to be obtained by the dealers in real time, so that they could serve the customers better. The visual-based shelf commodity detection methods have attracted extensive attention in recent years. Among them, the target detection method based on deep learning could automatically understand the picture features, which greatly promoted the development of goods shelf identification. This paper made a comparison and analysis between the common target detecting data set and the data set of goods on shelf from their features, advantages and application. It came up with a method to build an excellent data set of goods, and sorted out the data enhancement methods to perfect the data set for defective data sets. Then, from the three angles of large-scale package identification, small target identification and partial occluded recognition, the latest product package identification method was summarized. At last, the limitations of deep learning method in product identification were discussed and the future development direction was looked forward.

Deep Learning-based Image Recognition for Disaster Prevention Application

Deep Learning-based Image Recognition for Disaster Prevention Application