Transfer Learning Research Articles

Efficient fabric anomaly detection: A transfer learning framework with expedited training times

In industrial quality control, anomaly detection plays a critical role in identifying defective products. However, because of the rarity and time-consuming nature of defect collection, training models often rely solely on defect-free samples. This necessitates the use of unsupervised anomaly-detection techniques trained exclusively on defect-free data. Alternatively, defect data can be synthesized to augment the dataset with defective samples. In the textile industry, expeditious model training is crucial to ensure a smooth production flow. Unfortunately, most unsupervised methods require extensive training time. This paper proposes a novel transfer learning approach designed to achieve training times in seconds while effectively adapting the model to the target domain of fabric anomaly detection. The key contributions of our method include significantly reduced training times, up to 10 times faster than current state-of-the-art methods, and comparable performance in anomaly detection, achieving results on par with state-of-the-art approaches on benchmark datasets (MVTEC Anomaly Detection, TILDA, AITEX and DAGM). Additionally, our approach improves inference times, ensuring expedited and efficient anomaly detection during production. The proposed method offers a practical and efficient solution for real-time industrial quality control.

Transfer learning for crash design

When designing the structure of a new vehicle, car manufacturers need to ensure the compliance with strict safety requirements. Aiming to support the engineers in the early phase of this process, we propose a transfer learning framework for crashworthiness. This work explores the possibility to infer knowledge on future situations by exploiting data coming from past development processes. During the early phases of automotive development, assessing the crash safety implies dealing with the challenge of low data availability. Here, the engineers have no hardware test to rely on and can access only few finite element simulations. Under these circumstances, an attractive concept to investigate is the development of a machine learning approach able to learn from the past designs and to transfer the acquired knowledge to the new ones. Transfer learning can serve to this aim. With it, one learns the basic knowledge from a source domain A, and transfers it to a target domain B, characterized by low data availability. Here, we propose a transfer learning framework and apply it to an explicatory industrial crash example. The components produced in the past constitute the source domain; the new component design is the target domain. The proposed methodology can serve as an innovative solution to support car manufacturers in the early phase of vehicle development and thus improve the performance in crashworthiness scenarios.

An optimized heterogeneous multi-access edge computing framework based on transfer learning and artificial internet of things

In most practical applications, the feature space of the training datasets and the target domain datasets are inconsistent, or the data distribution between them is inconsistent, which leads to the problem of data starvation and makes it difficult for terminal devices to obtain high accurate results. Aiming at the problems of limited terminal device resources, low accuracy of data processing results, and unsatisfactory processing speed, a Heterogeneous Multi-access Edge Computing (MEC) Framework based on Transfer Learning (TL) is proposed, abbreviated as HMECF-TL. This framework adopts a cloud-edge-end three-layer architecture. It uses model transfer to optimize the Convolutional Neural Networks (CNN) model at each layer to achieve the goal of improving data processing speed and accuracy. Furthermore, a multi-agent Deep Reinforcement Learning Algorithm having Attention Mechanism (DRLAAM) is designed to further increase the timeliness performance of computation-intensive applications. The performance of HMECSF-TL framework is verified by simulation experiments, which not only reduces the delay by more than 24.66 %, but also improves the accuracy by more than 8.34 %. The framework not only increase the computing capacity to solve the shortage of terminal device resources, but also improve the quality of data processing to solve the problem of data starvation.

Utilizing deep learning models in an intelligent spiral drawing classification system for Parkinson's disease classification.

Parkinson's disease (PD) is a neurodegenerative illness that impairs normal human movement. The primary cause of PD is the deficiency of dopamine in the human brain. PD also leads to several other challenges, including insomnia, eating disturbances, excessive sleepiness, fluctuations in blood pressure, sexual dysfunction, and other issues. The suggested system is an extremely promising technological strategy that may help medical professionals provide accurate and unbiased disease diagnoses. This is accomplished by utilizing significant and unique traits taken from spiral drawings connected to Parkinson's disease. While PD cannot be cured, early administration of drugs may significantly improve the condition of a patient with PD. An expeditious and accurate clinical classification of PD ensures that efficacious therapeutic interventions can commence promptly, potentially impeding the advancement of the disease and enhancing the quality of life for both patients and their caregivers. Transfer learning models have been applied to diagnose PD by analyzing important and distinctive characteristics extracted from hand-drawn spirals. The studies were carried out in conjunction with a comparison analysis employing 102 spiral drawings. This work enhances current research by analyzing the effectiveness of transfer learning models, including VGG19, InceptionV3, ResNet50v2, and DenseNet169, for identifying PD using hand-drawn spirals. Transfer machine learning models demonstrate highly encouraging outcomes in providing a precise and reliable classification of PD. Actual results demonstrate that the InceptionV3 model achieved a high accuracy of 89% when learning from spiral drawing images and had a superior receiver operating characteristic (ROC) curve value of 95%. The comparison results suggest that PD identification using these models is currently at the forefront of PD research. The dataset will be enlarged, transfer learning strategies will be investigated, and the system's integration into a comprehensive Parkinson's monitoring and evaluation platform will be looked into as future research areas. The results of this study could lead to a better quality of life for Parkinson's sufferers, individualized treatment, and an early classification.

Hyperspectral Imaging Combined with Deep Transfer Learning to Evaluate Flavonoids Content in Ginkgo biloba Leaves.

Ginkgo biloba is a famous economic tree. Ginkgo leaves have been utilized as raw materials for medicines and health products due to their rich active ingredient composition, especially flavonoids. Since the routine measurement of total flavones is time-consuming and destructive, rapid, non-destructive detection of total flavones in ginkgo leaves is of significant importance to producers and consumers. Hyperspectral imaging technology is a rapid and non-destructive technique for determining the total flavonoid content. In this study, we discuss five modeling methods, and three spectral preprocessing methods are discussed. Bayesian Ridge (BR) and multiplicative scatter correction (MCS) were selected as the best model and the best pretreatment method, respectively. The spectral prediction results based on the BR + MCS treatment were very accurate (RTest2 = 0.87; RMSETest = 1.03 mg/g), showing a high correlation with the analytical measurements. In addition, we also found that the more and deeper the leaf cracks, the higher the flavonoid content, which helps to evaluate leaf quality more quickly and easily. In short, hyperspectral imaging is an effective technique for rapid and accurate determination of total flavonoids in ginkgo leaves and has great potential for developing an online quality detection system for ginkgo leaves.

Deep learning approach for detecting tomato flowers and buds in greenhouses on 3P2R gantry robot

In recent years, significant advancements have been made in the field of smart greenhouses, particularly in the application of computer vision and robotics for pollinating flowers. Robotic pollination offers several benefits, including reduced labor requirements and preservation of costly pollen through artificial tomato pollination. However, previous studies have primarily focused on the labeling and detection of tomato flowers alone. Therefore, the objective of this study was to develop a comprehensive methodology for simultaneously labeling, training, and detecting tomato flowers specifically tailored for robotic pollination. To achieve this, transfer learning techniques were employed using well-known models, namely YOLOv5 and the recently introduced YOLOv8, for tomato flower detection. The performance of both models was evaluated using the same image dataset, and a comparison was made based on their Average Precision (AP) scores to determine the superior model. The results indicated that YOLOv8 achieved a higher mean AP (mAP) of 92.6% in tomato flower and bud detection, outperforming YOLOv5 with 91.2%. Notably, YOLOv8 also demonstrated an inference speed of 0.7 ms when considering an image size of 1920×1080\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$1920 \ imes 1080$$\\end{document} pixels resized to 640×640\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$640 \ imes 640$$\\end{document} pixels during detection. The image dataset was acquired during both morning and evening periods to minimize the impact of lighting conditions on the detection model. These findings highlight the potential of YOLOv8 for real-time detection of tomato flowers and buds, enabling further estimation of flower blooming peaks and facilitating robotic pollination. In the context of robotic pollination, the study also focuses on the deployment of the proposed detection model on the 3P2R gantry robot. The study introduces a kinematic model and a modified circuit for the gantry robot. The position-based visual servoing method is employed to approach the detected flower during the pollination process. The effectiveness of the proposed visual servoing approach is validated in both un-clustered and clustered plant environments in the laboratory setting. Additionally, this study provides valuable theoretical and practical insights for specialists in the field of greenhouse systems, particularly in the design of flower detection algorithms using computer vision and its deployment in robotic systems used in greenhouses.

Radiogenomics based survival prediction of small-sample glioblastoma patients by multi-task DFFSP model.

In this paper, the multi-task dense-feature-fusion survival prediction (DFFSP) model is proposed to predict the three-year survival for glioblastoma (GBM) patients based on radiogenomics data. The contrast-enhanced T1-weighted (T1w) image, T2-weighted (T2w) image and copy number variation (CNV) is used as the input of the three branches of the DFFSP model. This model uses two image extraction modules consisting of residual blocks and one dense feature fusion module to make multi-scale fusion of T1w and T2w image features as backbone. Also, a gene feature extraction module is used to adaptively weight CNV fragments. Besides, a transfer learning module is introduced to solve the small sample problem and an image reconstruction module is adopted to make the model anatomy-aware under a multi-task framework. 256 sample pairs (T1w and corresponding T2w MRI slices) and 187 CNVs of 74 patients were used. The experimental results show that the proposed model can predict the three-year survival of GBM patients with the accuracy of 89.1 %, which is improved by 3.2 and 4.7 % compared with the model without genes and the model using last fusion strategy, respectively. This model could also classify the patients into high-risk and low-risk groups, which will effectively assist doctors in diagnosing GBM patients.

Beyond PhacoTrainer: Deep Learning for Enhanced Trabecular Meshwork Detection in MIGS Videos.

The purpose of this study was to develop deep learning models for surgical video analysis, capable of identifying minimally invasive glaucoma surgery (MIGS) and locating the trabecular meshwork (TM). For classification of surgical steps, we had 313 video files (265 for cataract surgery and 48 for MIGS procedures), and for TM segmentation, we had 1743 frames (1110 for TM and 633 for no TM). We used transfer learning to update a classification model pretrained to recognize standard cataract surgical steps, enabling it to also identify MIGS procedures. For TM localization, we developed three different models: U-Net, Y-Net, and Cascaded. Segmentation accuracy for TM was measured by calculating the average pixel error between the predicted and ground truth TM locations. Using transfer learning, we developed a model which achieved 87% accuracy for MIGS frame classification, with area under the receiver operating characteristic curve (AUROC) of 0.99. This model maintained a 79% accuracy for identifying 14 standard cataract surgery steps. The overall micro-averaged AUROC was 0.98. The U-Net model excelled in TM segmentation with an Intersection over union (IoU) score of 0.9988 and an average pixel error of 1.47. Building on prior work developing computer vision models for cataract surgical video, we developed models that recognize MIGS procedures and precisely localize the TM with superior performance. Our work demonstrates the potential of transfer learning for extending our computer vision models to new surgeries without the need for extensive additional data collection. Computer vision models in surgical videos can underpin the development of systems offering automated feedback for trainees, improving surgical training and patient care.

Optimizing Glaucoma Diagnosis with Deep Learning-Based Segmentation and Classification of Retinal Images

Glaucoma, a leading cause of permanent blindness worldwide, necessitates early detection to prevent vision loss, a task that is challenging and time-consuming when performed manually. This study proposes an automatic glaucoma detection method on enhanced retinal images using deep learning. The system analyzes retinal images, generating masks for the optic disc and optic cup, and providing a classification for glaucoma diagnosis. We employ a U-Net architecture with a pretrained residual neural network (ResNet34) for segmentation and an EfficientNetB0 for classification. The proposed framework is tested on publicly available datasets, including ORIGA, REFUGE, RIM-ONE DL, and HRF. Our work evaluated the U-Net model with five pretrained backbones (ResNet34, ResNet50, VGG19, DenseNet121, and EfficientNetB0) and examined preprocessing effects. We optimized model training with limited data using transfer learning and data augmentation techniques. The segmentation model achieves a mean intersection over union (mIoU) value of 0.98. The classification model shows remarkable performance with 99.9% training and 100% testing accuracy on ORIGA, 99.9% training and 99% testing accuracy on RIM-ONE DL, and 98% training and 100% testing accuracy on HRF. The proposed model outperforms related works and demonstrates potential for accurate glaucoma classification and detection tasks.

Turbidity assessment in coastal regions combining machine learning, numerical modeling, and remote sensing

ABSTRACT Machine learning models for water quality prediction often face challenges due to insufficient data and uneven spatial-temporal distributions. To address these issues, we introduce a framework combining machine learning, numerical modeling, and remote sensing imagery to predict coastal water turbidity, a key water quality proxy. This approach was tested in the Great Lakes region, specifically Cleveland Harbor, Lake Erie. We trained models using observed and synthetic data from 3D numerical models and tested them against in situ and remote sensing data from PlanetLabs' Dove satellites. High-resolution (HR) data improved prediction accuracy, with RMSE values of 0.154 and 0.146 log10(FNU) and R2 values of 0.92 and 0.93 for validation and test datasets, respectively. Our study highlights the importance of unified turbidity measures for data comparability. The machine learning model demonstrated skill in predicting turbidity through transfer learning, indicating applicability in diverse, data-scarce regions. This approach can enhance decision support systems for coastal environments by providing accurate, timely predictions of water quality variables. Our methodology offers robust strategies for turbidity and water quality monitoring and holds significant potential for improving input data quality for numerical models and developing predictive models from remote sensing data.

NPSFF-Net: Enhanced Building Segmentation in Remote Sensing Images via Novel Pseudo-Siamese Feature Fusion

Building segmentation has extensive research value and application prospects in high-resolution remote sensing image (HRSI) processing. However, complex architectural contexts, varied building morphologies, and non-building occlusions make building segmentation challenging. Compared with traditional methods, deep learning-based methods present certain advantages in terms of accuracy and intelligence. At present, the most popular option is to first apply a single neural network to encode an HRSI, then perform a decoding process through up-sampling or using a transposed convolution operation, and then finally obtain the segmented building image with the help of a loss function. Although effective, this approach not only tends to lead to a loss of detail information, but also fails to fully utilize the contextual features. As an alternative, we propose a novel network called NPSFF-Net. First, using an improved pseudo-Siamese network composed of ResNet-34 and ResNet-50, two sets of deep semantic features of buildings are extracted with the support of transfer learning, and four encoded features at different scales are obtained after fusion. Then, information from the deepest encoded feature is enriched using a feature enhancement module, and the resolutions are recovered via the operations of skip connections and transposed convolutions. Finally, the discriminative features of buildings are obtained using the designed feature fusion algorithm, and the optimal segmentation model is obtained by fitting a cross-entropy loss function. Our method obtained intersection-over-union values of 89.45% for the Aerial Imagery Dataset, 71.88% for the Massachusetts Buildings Dataset, and 68.72% for the Satellite Dataset I.

Efficient Detection of Apparent Defects in Subway Tunnel Linings Based on Deep Learning Methods

High-precision and rapid detection of apparent defects in subway tunnel linings is crucial for ensuring the structural integrity of tunnels and the safety of train operations. However, current methods often do not adequately account for the spatial characteristics of these defects and perform poorly in detecting and extracting small-scale defects, which limits the accuracy of detection and geometric parameter extraction. To address these challenges, this paper proposes an efficient algorithm for detecting and extracting apparent defects in subway tunnels. Firstly, YOLOv8 was selected as the foundational architecture due to its comprehensive performance. The coordinate attention module and Bottleneck Transformer 3 were then integrated into the model’s backbone to enhance the focus on defect-prone areas and improve the learning of feature relationships between defects and other infrastructure. Subsequently, a high-resolution detection layer was added to the model’s head to further improve sensitivity to subtle defects. Additionally, a low-quality crack dataset was created using an open access dataset, and transfer learning combined with Real-ESRGAN was employed to enhance the detail and resolution of fine cracks. The results of the field experiments demonstrate that the proposed model significantly improves detection accuracy in high-incidence areas and for small-scale defects, achieving a mean average precision (mAP) of 87% in detecting cracks, leakage, exfoliation, and related infrastructure defects. Furthermore, the crack enhancement techniques substantially improve the representation of fine-crack details, increasing feature extraction accuracy by a factor of four. The findings of this paper could provide crucial technical support for the automated operation and maintenance of metro tunnels.

Detection of malignant breast tissue using SAR observation with microwave imaging and convolutional neural network

The research paper introduces a deep learning approach for distinguishing between benign and malignant breast tissues using Ultra Wide Band (UWB) microwave imaging. A multi-resonant monopole microstrip patch antenna was crafted and placed on a female breast phantom model containing a tumor. The phantom’s dielectric properties were adjusted to mimic those of benign and malignant tissues. The Specific Absorption Rate (SAR) was analyzed by rotating the antenna at various angles throughout the phantom’s trajectory. Images capturing the SAR distribution over the breast phantom at different antenna resonances were obtained. The popular Convolutional Neural Network (CNN) based AlexNet model was employed for autonomous feature learning from SAR images. To address the challenge of overfitting with a limited dataset, the study utilized image augmentation and transfer learning methods. The proposed model achieved an impressive 98.59% accuracy in classifying benign and malignant breast phantom images, surpassing the performance of three other CNN models—VggNet16, GoogleNet, and ResNet. Notably, this microwave imaging system based on SAR images detected tumors as small as 1 mm with an accuracy of 97.08% outperforming existing malignancy screening techniques.

A Physical‐Informed Neural Network for Improving Air‐Sea Turbulent Heat Flux Parameterization

AbstractThe parameterizations of air‐sea turbulent heat flux are one of the major bottlenecks in atmosphere‐ocean coupled model development, which play a crucial role in sea surface temperature (SST) prediction. Recently, neural networks start to be applied for the development of parameterizations of interface turbulent heat flux. However, these new parameterizations are primairily developed for specific regions and have not been tested in real atmosphere‐ocean coupled models. In this study, we propose a new air‐sea heat flux parameterization using a physical‐informed neural network (PINN) based on multiple observational data sets worldwide. Evaluated with an independent observation data set, it is shown that the PINN can significantly reduce the RMSE of latent heat flux by at least about 48.6% compared to three traditional bulk formulas. Moreover, the PINN can be flexibly updated with new observational data by transfer learning. To test the performance of the new parameterization in realistic application, we implement the PINN into a global ocean‐atmosphere coupled model and make seasonal forecasts for the first time. The PINN markedly reduce the errors of equatorial SST forecast, indicating a good performance of the PINN‐based air‐sea turbulent heat flux scheme. Noticeably, due to limited observational data, the NN‐based parameterizations tend to underestimate heat flux at high wind speeds compared with bulk formula‐based parameterizations. With more data available at extreme conditions, the PINN can be improved via transfer learning and need to be futher evaluated. This study suggests that PINN‐based air‐sea heat flux parameterization is promising to improve SST simulation.

Effect of different weight initialization strategies on transfer learning for plant disease detection

AbstractThe weight initialization technique for transfer learning refers to the practice of using pretrained models that can be modified to solve new problems, instead of starting the training process from scratch. In this study, six different transfer learning weight initialization strategies were proposed for plant disease detection: scratch (i.e., random initialization), pretrained model on cross‐domain (ImageNet), model trained on related domain (ISIC 2019), model trained on related domain (ISIC 2019) with cross‐domain (ImageNet) weights, model trained on same domain (PlantVillage), and model trained on same domain (PlantVillage) with cross‐domain weights (ImageNet). Weights from each strategy were transferred to a target dataset (Plant Pathology 2021). These strategies were implemented using eight deep learning architectures. It was observed that transferring from any strategy led to an average acceleration of convergence ranging from 33.88% to 73.16% in mean loss and an improvement of 8.72%–42.12% in mean F1‐score compared to the scratch strategy. Moreover, although smaller and less comprehensive than ImageNet, transferring information from the same domain or related domain proved to be competitive compared to transferring from ImageNet. This indicates that ImageNet, which is widely favoured in the literature, may not necessarily represent the most optimal transfer source for the given context. In addition, to identify which strategies have significant differences, a post hoc analysis using Tukey's HSD test was conducted. Finally, the classifications made by the proposed models were visualized using Grad‐CAM to provide a qualitative understanding of how different weight initialization strategies affect the focus areas of the models.

Distributed and Collaborative Lightweight Edge Federated Learning for IoT Zombie Devices Detection

The fast development of artificial intelligence and Internet of Things (IoT) technologies has enabled various applications of smart cities, e.g., smart monitoring and surveillance. However, vulnerabilities of IoT devices bring new threats to the security of smart cities. To identify ubiquitous IoT botnet attacks, a distributed and collaborative lightweight edge federated learning model for IoT zombie devices detection is proposed, named FIOT. To reduce computational complexity and enhance the adaptability to new attack environment at the network edge, FIOT is designed in a lightweight manner based on feature dimensionality reduction and transfer learning. Three IoT botnet datasets are used to validate the effectiveness of the proposed FIOT. Experimental results show that FIOT has an accuracy loss of less than 3% in terms of F1 value compared to the centralized learning, but the training time of FIOT is only 14.3% of that of centralized learning. While ensuring high detection accuracy, the number of parameters of FIOT is compressed to 37.58% of the comparison method.

Joint estimation for multisource Gaussian graphical models based on transfer learning

This study considers data from multiple sources for Gaussian graphical models with one target graph and several auxiliary graphs. We propose a method called joint estimation for multisource Gaussian graphical models (JEM-GGM) to achieve a stable and accurate estimate of the target graph. Using the information from the auxiliary graphs, the proposed method is used to effectively solve the problem of small sample sizes. In this method, equivalent regression models are developed for graphs and the difference between the auxiliary and target graphs was penalized to ensure computational efficiency and improve estimation accuracy. Simulations revealed that the proposed method always outperformed other methods in terms of estimation and prediction accuracy. The application of this method to breast and lymphatic cancer cell lines reveals that the proposed method always obtains a sparse collection of important genome pairs.

Simultaneously predicting SPAD and water content in rice leaves using hyperspectral imaging with deep multi-task regression and transfer component analysis.

Water content and chlorophyll content are important indicators for monitoring rice growth status. Simultaneous detection of water content and chlorophyll content is of significance. Different varieties of rice show differences in phenotype, resulting in the difficulties of establishing a universal model. In this study, hyperspectral imaging was used to detect the Soil and Plant Analyzer Development (SPAD) values and water content of fresh rice leaves of three rice varieties (Jiahua 1, Xiushui 121 and Xiushui 134). Both partial least squares regression and convolutional neural networks were used to establish single-task and multi-task models. Transfer component analysis (TCA) was used as transfer learning to learn the common features to achieve an approximate identical distribution between any two varieties. Single-task and multi-task models were also built using the features of the source domain, and these models were applied to the target domain. These results indicated that for models of each rice variety the prediction accuracy of most multi-task models was close to that of single-task models. As for TCA, the results showed that the single-task model achieved good performance for all transfer learning tasks. Compared with the original model, good and differentiated results were obtained for the models using features learned by TCA for both the source domain and target domain. The multi-task models could be constructed to predict SPAD values and water content simultaneously and then transferred to another rice variety, which could improve the efficiency of model construction and realize rapid detection of rice growth indicators. © 2024 Society of Chemical Industry.

Cotton-YOLO-Seg: An Enhanced YOLOV8 Model for Impurity Rate Detection in Machine-Picked Seed Cotton

The detection of the impurity rate in machine-picked seed cotton is crucial for precision agriculture. This study proposes a novel Cotton-YOLO-Seg cotton-impurity instance segmentation algorithm based on the you only look once version 8 small segmentation model (Yolov8s-Seg). The algorithm achieves precise pixel-level segmentation of cotton and impurities in seed cotton images and establishes a detection model for the impurity rate, enabling accurate detection of the impurity rate in machine-picked cotton. The proposed algorithm removes the Pyramid 4 (P4) feature layer and incorporates Multi-Scale Convolutional Block Attention (MSCBCA) that integrates the Convolutional Block Attention Module (CBAM) and Multi-Scale Convolutional Attention (MSCA) into the Faster Implementation of Cross Stage Partial Bottleneck with 2 Convolutions (C2f) module of the feature extraction network, forming a novel C2f_MSCBCA module. The SlimNeck structure is introduced in the feature fusion network by replacing the P4 feature layer with the small-target detection layer Pyramid 2 (P2). Additionally, transfer learning is employed using the Common Objects in Context (COCO) instance segmentation dataset. The analysis of 100 groups of cotton image samples shows that the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) for impurity rate detection are 0.29%, 0.33%, and 3.70%, respectively, which are reduced by 52.46%, 48.44%, and 53.75% compared to the Yolov8s-seg model. The Precision (P), Recall (R), and mean Average Precision at an intersection over union of 0.5 (mAP@0.5) are 85.4%, 78.4%, and 80.8%, respectively, which are improved by 4.2%, 6.2%, and 6.4% compared to Yolov8s-seg model, significantly enhancing the segmentation performance of minor impurities. The Cotton-YOLO-Seg model demonstrates practical significance for precisely detecting the impurity rate in machine-picked seed cotton.

Dual-Path Neural Network based on Deep Features and Transfer Learning for Bearing Fault Diagnosis

The safe operation of mechanical equipment is a basic requirement in the industrial manufacturing process. As an important component of mechanical equipment, rolling bearings are prone to unforeseen failures due to long-term operation under complex conditions. The occurrence of failures, no matter how big or small, will cause economic losses. Therefore, reliable diagnosis of rolling bearings is crucial. Aiming at the problem of insufficient feature recognition of convolutional neural networks under strong noise background, a deep feature extraction network integrating multiscale convolutional neural network (MSCNN) and bidirectional gated recurrent unit (BiGRU) is proposed. MSCNN and BiGRU are used to extract multiscale features and temporal features from noisy vibration signals respectively, and different weights are assigned to the fused features through the attention mechanism module to achieve important feature selection. Furthermore, in order to solve the problem of different feature distributions between the source domain and the target domain under variable working conditions, transfer learning is introduced in the proposed deep feature extraction network. The difference in feature distribution between the source domain and the target domain is measured by a multi-level distance formula, and the measurement result is added to the loss function. The back propagation of the loss is used to achieve the alignment of feature distribution between the source domain and the target domain. Finally, the model uses the SoftMax function as a classifier for rolling bearing fault diagnosis. Experimental comparison and analysis show that the proposed model has good migration ability and achieves a higher fault identification accuracy.