How to discriminate wood of CITES-listed tree species from their look-alikes: using an attention mechanism with the ResNet model on an enhanced macroscopic image dataset.

Introduction: Global illegal trade in timbers is a major cause of the loss of tree species diversity. The Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) has been developed to combat the illegal international timber trade. Its implementation relies on accurate wood identification techniques for field screening. However, meeting the demand for timber field screening at the species level using the traditional wood identification method depending on wood anatomy is complicated, time-consuming, and challenging for enforcement officials who did not major in wood science. Methods: This study constructed a CITES-28 macroscopic image dataset, including 9,437 original images of 279 xylarium wood specimens from 14 CITES-listed commonly traded tree species and 14 look-alike species. We evaluated a suitable wood image preprocessing method and developed a highly effective computer vision classification model, SE-ResNet, on the enhanced image dataset. The model incorporated attention mechanism modules [squeeze-and-excitation networks (SENet)] into a convolutional neural network (ResNet) to identify 28 wood species. Results: The results showed that the SE-ResNet model achieved a remarkable 99.65% accuracy. Additionally, image cropping and rotation were proven effective image preprocessing methods for data enhancement. This study also conducted real-world identification using images of new specimens from the timber market to test the model and achieved 82.3% accuracy. Conclusion: This study presents a convolutional neural network model coupled with the SENet module to discriminate CITES-listed species with their look-alikes and investigates a standard guideline for enhancing wood transverse image data, providing a practical computer vision method tool to protect endangered tree species and highlighting its substantial potential for CITES implementation.

Crash severity analysis: A data-enhanced double layer stacking model using semantic understanding.

The crash severity analysis is of significant importance in traffic crash prevention and emergency resource allocation. A range of innovations offers potential traffic crash severity prediction models to improve road safety. However, the semantic information inherent in traffic crash data, which is crucial in enabling a deeper understanding of its underlying factors and impacts, has yet to be fully utilized. Moreover, traffic crash data are commonly characterized by a small sample size, which leads to sample imbalance problem resulting in prediction performance decline. To tackle these problems, we propose a semantic understanding-based data-enhanced double-layer stacking model, named EnLKtreeGBDT, for crash severity prediction. Specifically, to fully leverage the inherent semantic information within traffic crash data and analyze the factors influencing crashes, we design a semantic enhancement module for multi-dimensional feature extraction. This module aims to enhance the understanding of crash semantics and improve prediction accuracy. Then we introduce a data enhancement module that utilizes data denoising and migration techniques to address the challenge of data imbalance, reducing the prediction model's dependence on large sample crash data. Furthermore, we construct a two-layer stacking model that combines multiple linear and nonlinear classifiers. This model is designed to augment the capability of learning linear and nonlinear mixed relationships, thereby improving the accuracy of predicting the severity of crashes on complex urban roads. Experiments on historical datasets of UK road safety crashes validate the effectiveness of the proposed model, and superior performance of prediction precision is achieved compared with the state-of-the-arts. The ablation experiments on both semantic and data enhancement modules further confirm the indispensability of each module in the proposed model.

Rail surface defect data enhancement method based on improved ACGAN.

Rail surface defects present a significant safety concern in railway operations. However, the scarcity of data poses challenges for employing deep learning in defect detection. This study proposes an enhanced ACGAN augmentation method to address these issues. Residual blocks mitigate vanishing gradient problems, while a spectral norm regularization-constrained discriminator improves stability and image quality. Substituting the generator's deconvolution layer with upsampling and convolution operations enhances computational efficiency. A gradient penalty mechanism based on regret values addresses gradient abnormality concerns. Experimental validation demonstrates superior image clarity and classification accuracy compared to ACGAN, with a 17.6% reduction in FID value. MNIST dataset experiments verify the model's generalization ability. This approach offers practical value for real-world applications.

A generic optimization-based enhancement method for trajectory data: Two plus one.

Trajectory data play a vital role in the field of traffic research such as vehicle safety, traffic flow, and intelligent vehicles. The quality of trajectory data will determine the safety effectiveness of both research and practical applications. Effectively filtering out noise and errors from trajectory data is crucial for improving data quality and further research. However, most enhancement methods only focus on the smoothness of trajectory but overlook abrupt changes. The processed trajectory still exist issues such as incomplete elimination of inconsistency and loss of driving characteristics. In this paper, we propose a generic optimization-based enhancement method to address the issues above. We propose a bilevel optimization method combined with ℓl1 and ℓl2 trend filter. First, we design a lℓ2 trend filter to fuse raw trajectory data and eliminate the inconsistency. Next, we utilize the lℓ1 trend filter to optimize the data, ensuring physical feasibility and preserving abrupt changes (emergency driving characteristics). Then, we validate the effectiveness of the method through evaluation metrics and prediction models. The generic optimization-based enhancement method proposed in this paper ensures the safety of both research and application by providing high-quality trajectory data.

ECG classification efficient modeling with artificial bee colony optimization data augmentation and attention mechanism.

In addressing the key issues of the data imbalance within ECG signals and modeling optimization, we employed the TimeGAN network and a local attention mechanism based on the artificial bee colony optimization algorithm to enhance the performance and accuracy of ECG modeling. Initially, the TimeGAN network was introduced to rectify data imbalance and create a balanced dataset. Furthermore, the artificial bee colony algorithm autonomously searched hyperparameter configurations by minimizing Wasserstein distance. Control experiments revealed that data augmentation significantly boosted classification accuracy to 99.51%, effectively addressing challenges with unbalanced datasets. Moreover, to overcome bottlenecks in the existing network, the introduction of the Efficient network was adopted to enhance the performance of modeling optimized with attention mechanisms. Experimental results demonstrated that this integrated approach achieved an impressive overall accuracy of 99.70% and an average positive prediction rate of 99.44%, successfully addressing challenges in ECG signal identification, classification, and diagnosis.

Data Fusion of RGB and Depth Data with Image Enhancement.

Since 3D sensors became popular, imaged depth data are easier to obtain in the consumer sector. In applications such as defect localization on industrial objects or mass/volume estimation, precise depth data is important and, thus, benefits from the usage of multiple information sources. However, a combination of RGB images and depth images can not only improve our understanding of objects, capacitating one to gain more information about objects but also enhance data quality. Combining different camera systems using data fusion can enable higher quality data since disadvantages can be compensated. Data fusion itself consists of data preparation and data registration. A challenge in data fusion is the different resolutions of sensors. Therefore, up- and downsampling algorithms are needed. This paper compares multiple up- and downsampling methods, such as different direct interpolation methods, joint bilateral upsampling (JBU), and Markov random fields (MRFs), in terms of their potential to create RGB-D images and improve the quality of depth information. In contrast to the literature in which imaging systems are adjusted to acquire the data of the same section simultaneously, the laboratory setup in this study was based on conveyor-based optical sorting processes, and therefore, the data were acquired at different time periods and different spatial locations. Data assignment and data cropping were necessary. In order to evaluate the results, root mean square error (RMSE), signal-to-noise ratio (SNR), correlation (CORR), universal quality index (UQI), and the contour offset are monitored. With JBU outperforming the other upsampling methods, achieving a meanRMSE = 25.22, mean SNR = 32.80, mean CORR = 0.99, and mean UQI = 0.97.

Few-shot learning based on deep learning: A survey.

In recent years, with the development of science and technology, powerful computing devices have been constantly developing. As an important foundation, deep learning (DL) technology has achieved many successes in multiple fields. In addition, the success of deep learning also relies on the support of large-scale datasets, which can provide models with a variety of images. The rich information in these images can help the model learn more about various categories of images, thereby improving the classification performance and generalization ability of the model. However, in real application scenarios, it may be difficult for most tasks to collect a large number of images or enough images for model training, which also restricts the performance of the trained model to a certain extent. Therefore, how to use limited samples to train the model with high performance becomes key. In order to improve this problem, the few-shot learning (FSL) strategy is proposed, which aims to obtain a model with strong performance through a small amount of data. Therefore, FSL can play its advantages in some real scene tasks where a large number of training data cannot be obtained. In this review, we will mainly introduce the FSL methods for image classification based on DL, which are mainly divided into four categories: methods based on data enhancement, metric learning, meta-learning and adding other tasks. First, we introduce some classic and advanced FSL methods in the order of categories. Second, we introduce some datasets that are often used to test the performance of FSL methods and the performance of some classical and advanced FSL methods on two common datasets. Finally, we discuss the current challenges and future prospects in this field.

Bionic study of distance-azimuth discrimination of multi-scattered point objects in bat bio-sonar.

This paper presents a novel approach to enhance the discrimination capacity of multi-scattered point objects in bat bio-sonar. A broadband interferometer mathematical model is developed, incorporating both distance and azimuth information, to simulate the transmitted and received signals of bats. The Fourier transform is employed to simulate the preprocessing step of bat information for feature extraction. Furthermore, the bat bio-sonar model based on convolutional neural network (BS-CNN) is constructed to compensate for the limitations of conventional machine learning and CNN networks, including three strategies: Mix-up data enhancement, joint feature and hybrid atrous convolution module. The proposed BS-CNN model emulates the perceptual nerves of the bat brain for distance-azimuth discrimination and compares with four conventional classifiers to assess its discrimination efficacy. Experimental results demonstrate that the overall discrimination accuracy of the BS-CNN model is 93.4%, surpassing conventional CNN networks and machine learning methods by at least 5.9%. This improvement validates the efficacy of the BS-CNN bionic model in enhancing the discrimination accuracy in bat bio-sonar and offers valuable references for radar and sonar target classification.

ECG data enhancement method using generate adversarial networks based on Bi-LSTM and CBAM.

Objective.The classification performance of electrocardiogram (ECG) classification algorithms is easily affected by data imbalance, which often leads to poor model prediction performance for a few classes and a consequent decrease in the overall performance of the model.Approach.To address this problem, this paper proposed an ECG data augmentation method based on a generative adversarial network (GAN) that combines bidirectional long short-term memory (Bi-LSTM) networks and convolutional block attention mechanism (CBAM) to improve the overall performance of ECG classification models. In this paper, we used two ECG databases, namely the MIT-BIH arrhythmia (MIT-BIH-AR) database and the Chinese cardiovascular disease database (CCDD). The quality of the ECG signals produced by the generated models was assessed using the percent relative difference, root mean square error, Frechet distance, dynamic time warping (DTW), and Pearson correlation metrics. In addition, we also validated the impact of our proposed data augmentation method on ECG classification performance on MIT-BIH-AR database and CCDD.Main results.On the MIT-BIH-AR database, the overall accuracy of the data-enhanced balanced dataset was improved to 99.46% for 15 types of heartbeat classification task. On the CCDD, which focuses on the detection of ventricular precession (PVC), the overall accuracy of PVC detection improved to 99.15% after performing data enhancement.Significance.The experimental results indicate that the data augmentation method proposed in this paper can further improve the ECG classification performance.

Deep learning-based ultra-fast identification of Raman spectra with low signal-to-noise ratio.

Ensuring the correct use of cell lines is crucial to obtaining reliable experimental results and avoiding unnecessary waste of resources. Raman spectroscopy has been confirmed to be able to identify cell lines, but the collection time is usually 10-30 s. In this study, we acquired Raman spectra of five cell lines with integration times of 0.1 and 8 s, respectively, and the average accuracy of using long-short memory neural network to identify the spectra of 0.1 s was 95%, and the average accuracy of identifying the spectra of 8 s was 99.8%. At the same time, we performed data enhancement of 0.1 s spectral data by real-valued non-volume preserving method, and the recognition average accuracy of long-short memory neural networks recognition of the enhanced spectral data was improved to 96.2%. With this method, we shorten the acquisition time of Raman spectra to 1/80 of the original one, which greatly improves the efficiency of cell identification.

[A study on the application of cross-frequency coupling characteristics of neural oscillation in the diagnosis of mild cognitive impairment].

In order to fully explore the neural oscillatory coupling characteristics of patients with mild cognitive impairment (MCI), this paper analyzed and compared the strength of the coupling characteristics for 28 MCI patients and 21 normal subjects under six different-frequency combinations. The results showed that the difference in the global phase synchronization index of cross-frequency coupling under δ-θ rhythm combination was statistically significant in the MCI group compared with the normal control group ( P = 0.025, d = 0.398). To further validate this coupling feature, this paper proposed an optimized convolutional neural network model that incorporated a time-frequency data enhancement module and batch normalization layers to prevent overfitting while enhancing the robustness of the model. Based on this optimized model, with the phase locking value matrix of δ-θ rhythm combination as the single input feature, the diagnostic accuracy of MCI patients was (95.49 ± 4.15)%, sensitivity and specificity were (93.71 ± 7.21)% and (97.50 ± 5.34)%, respectively. The results showed that the characteristics of the phase locking value matrix under the combination of δ-θ rhythms can adequately reflect the cognitive status of MCI patients, which is helpful to assist the diagnosis of MCI.

A novel wavelet-transform-based convolution classification network for cervical lymph node metastasis of papillary thyroid carcinoma in ultrasound images.

Preoperative assessment of cervical lymph nodes metastasis (CLNM) for accurate qualitative and locating diagnosis is important for choosing the best treatment option for patients with papillary thyroid cancer. Non-destructive, non-invasive ultrasound is currently the imaging method of choice for lymph node metastatic assessment. For lymph node characteristics and ultrasound images, this paper proposes a multitasking network framework for diagnosing metastatic lymph nodes in ultrasound images, in which localization module not only provides information on the location of lymph nodes to focus on the peripheral and self regions of lymph nodes, but also provides structural features of lymph nodes for subsequent classification module. In the classification module, we design a novel wavelet-transform-based convolution network. Wavelet transform is introduced into the deep learning convolution module to analyze ultrasound images in both spatial and frequency domains, which effectively enriches the feature information and improves the classification performance of the model without increasing the model parameters. We collected 510 patient data (N = 1376) from Shanghai Sixth People's Hospital regarding ultrasound lymph nodes in the neck, as well as used three publicly available ultrasound datasets, including SCUI2020 (N = 2914), DDTI (N = 480), and BUSI (N = 780). Compared to the optimal two-stage model, our model has improved its accuracy and AUC indexes by 5.83% and 4%, which outperforms the two-stage architectures and also surpasses the latest classification networks.

A Novel Method for Filled/Unfilled Grain Classification Based on Structured Light Imaging and Improved PointNet+.

China is the largest producer and consumer of rice, and the classification of filled/unfilled rice grains is of great significance for rice breeding and genetic analysis. The traditional method for filled/unfilled rice grain identification was generally manual, which had the disadvantages of low efficiency, poor repeatability, and low precision. In this study, we have proposed a novel method for filled/unfilled grain classification based on structured light imaging and Improved PointNet++. Firstly, the 3D point cloud data of rice grains were obtained by structured light imaging. And then the specified processing algorithms were developed for the single grain segmentation, and data enhancement with normal vector. Finally, the PointNet++ network was improved by adding an additional Set Abstraction layer and combining the maximum pooling of normal vectors to realize filled/unfilled rice grain point cloud classification. To verify the model performance, the Improved PointNet++ was compared with six machine learning methods, PointNet and PointConv. The results showed that the optimal machine learning model is XGboost, with a classification accuracy of 91.99%, while the classification accuracy of Improved PointNet++ was 98.50% outperforming the PointNet 93.75% and PointConv 92.25%. In conclusion, this study has demonstrated a novel and effective method for filled/unfilled grain recognition.

Track Fastener Defect Detection Model Based on Improved YOLOv5s.

Defect detection of track fasteners is a prerequisite for safe and reliable railroad operation. The traditional manual visual inspection method has been unable to meet the growing demand for railroad network inspection in China. To achieve the need for accurate, fast, and intelligent detection of rail fasteners, this paper proposes a rail fastener defect detection model based on improved YOLOv5s. Firstly, the convolutional block attention module (CBAM) is added to the Neck network of the YOLOv5s model to enhance the extraction of essential features by the model and suppress the information of minor features. Secondly, a weighted bidirectional feature pyramid network (BiFPN) is introduced to realize the multi-scale feature fusion of the model. Finally, the K-means++ algorithm is used to re-cluster the dataset to obtain the anchor box suitable for the fastener dataset and improve the positioning ability of the model. The experimental results show that the improved model achieves an average mean precision (mAP) of 97.4%, a detection speed of 27.3 FPS, and a model memory occupancy of 15.5 M. Compared with the existing target detection model, the improved model has the advantages of high detection accuracy, fast detection speed, and small model memory occupation, which can provide technical support for edge deployment of rail fastener defect detection.

OSCA-finder: Redefining the assay of kidney disease diagnostic through metabolomics and deep learning.

Liquid chromatography-mass spectrometry (LC-MS) is a platform for urine and blood sample analysis. However, the high variability in the urine sample reduced the confidence of metabolite identification. Therefore, pre and post-calibration operations are inevitable to ensure an accurate urine biomarker analysis. In this study, the phenomenon of a higher creatinine concentration variable in ureteropelvic junction obstruction (UPJO) patient urine samples than in healthy people was revealed, indicating the urine biomarker discovery of UPJO patients is not adapted to the creatinine calibrate strategy. Therefore, we proposed a pipeline "OSCA-Finder" to reshape the urine biomarker analysis. First, to ensure a more stable peak shape and total ion chromatography, we applied the product of osmotic pressure and injection volume as a calibration principle and integrated it with an online mixer dilution. Therefore, we obtained the most peaks and identified more metabolites in a urine sample with peak area group CV<30%. A data-enhanced strategy was applied to reduce the overfit while training a neural network binary classifier with an accuracy of 99.9%. Finally, seven accurate urine biomarkers combined with a binary classifier were applied to distinguish UPJO patients from healthy people. The results show that the UPJO diagnostic strategy based on urine osmotic pressure calibration has more potential than ordinary strategies.

Rice pest identification based on multi-scale double-branch GAN-ResNet.

Rice production is crucial to the food security of all human beings, and how rice pests and diseases can be effectively prevented in and timely detected is a hotspot issue in the field of smart agriculture. Deep learning has become the preferred method for rice pest identification due to its excellent performance, especially in the aspect of autonomous learning of image features. However, in the natural environment, the dataset is too small and vulnerable to the complex background, which easily leads to problems such as overfitting, and too difficult to extract the fine features during the process of training. To solve the above problems, a Multi-Scale Dual-branch structural rice pest identification model based on a generative adversarial network and improved ResNet was proposed. Based on the ResNet model, the ConvNeXt residual block was introduced to optimize the calculation ratio of the residual blocks, and the double-branch structure was constructed to extract disease features of different sizes in the input disease images, which it adjusts the size of the convolution kernel of each branch. In the complex natural environment, data pre-processing methods such as random brightness and motion blur, and data enhancement methods such as mirroring, cropping, and scaling were used to allow the dataset of 5,932 rice disease images captured from the natural environment to be expanded to 20,000 by the dataset in this paper. The new model was trained on the new dataset to identify four common rice diseases. The experimental results showed that the recognition accuracy of the new rice pest recognition model, which was proposed for the first time, improved by 2.66% compared with the original ResNet model. Under the same experimental conditions, the new model had the best performance when compared with classical networks such as AlexNet, VGG, DenseNet, ResNet, and Transformer, and its recognition accuracy could be as high as 99.34%. The model has good generalization ability and excellent robustness, which solves the current problems in rice pest identification, such as the data set is too small and easy to lead to overfitting, and the picture background is difficult to extract disease features, and greatly improves the recognition accuracy of the model by using a multi-scale double branch structure. It provides a superior solution for crop pest and disease identification.

Hybrid Granularities Transformer for Fine-Grained Image Recognition.

Many current approaches for image classification concentrate solely on the most prominent features within an image, but in fine-grained image recognition, even subtle features can play a significant role in model classification. In addition, the large variations in the same class and small differences between different categories that are unique to fine-grained image recognition pose a great challenge for the model to extract discriminative features between different categories. Therefore, we aim to present two lightweight modules to help the network discover more detailed information in this paper. (1) Patches Hidden Integrator (PHI) module randomly selects patches from images and replaces them with patches from other images of the same class. It allows the network to glean diverse discriminative region information and prevent over-reliance on a single feature, which can lead to misclassification. Additionally, it does not increase the training time. (2) Consistency Feature Learning (CFL) aggregates patch tokens from the last layer, mining local feature information and fusing it with the class token for classification. CFL also utilizes inconsistency loss to force the network to learn common features in both tokens, thereby guiding the network to focus on salient regions. We conducted experiments on three datasets, CUB-200-2011, Stanford Dogs, and Oxford 102 Flowers. We achieved experimental results of 91.6%, 92.7%, and 99.5%, respectively, achieving a competitive performance compared to other works.

ST-CenterNet: Small Target Detection Algorithm with Adaptive Data Enhancement.

General target detection with deep learning has made tremendous strides in the past few years. However, small target detection sometimes is associated with insufficient sample size and difficulty in extracting complete feature information. For safety during autonomous driving, remote signs and pedestrians need to be detected from driving scenes photographed by car cameras. In the early period of a medical lesion, because of the small area of the lesion, target detection is of great significance to detect masses and tumors for accurate diagnosis and treatment. To deal with these problems, we propose a novel deep learning model, named CenterNet for small targets (ST-CenterNet). First of all, due to the lack of visual information on small targets in the dataset, we extracted less discriminative features. To overcome this shortcoming, the proposed selective small target replication algorithm (SSTRA) was used to realize increasing numbers of small targets by selectively oversampling them. In addition, the difficulty of extracting shallow semantic information for small targets results in incomplete target feature information. Consequently, we developed a target adaptation feature extraction module (TAFEM), which was used to conduct bottom-up and top-down bidirectional feature extraction by combining ResNet with the adaptive feature pyramid network (AFPN). The improved new network model, AFPN, was added to solve the problem of the original feature extraction module, which can only extract the last layer of the feature information. The experimental results demonstrate that the proposed method can accurately detect the small-scale image of distributed targets and simultaneously, at the pixel level, classify whether a subject is wearing a safety helmet. Compared with the detection effect of the original algorithm on the safety helmet wearing dataset (SHWD), we achieved mean average precision (mAP) of 89.06% and frames per second (FPS) of 28.96, an improvement of 18.08% mAP over the previous method.

Hardware Accelerated Design of a Dual-Mode Refocusing Algorithm for SAR Imaging Systems.

The use of satellite synthetic aperture radar (SAR) for moving target imaging has gained popularity recently. Researchers are focused on improving its imaging quality. To achieve high-quality and fast imaging, we have developed a dual-mode refocusing algorithm. We optimized the algorithm's target speed estimation and carried out data enhancement and quantization design for SAR image refocusing. The design is implemented on a Xilinx XC5VFX130T FPGA. The dual-mode image data are based on a slice size of 512 × 512 for slice mode and 256 × 256 for scan mode in a time-series function simulation. The serial-parallel conversion and pipeline design balances the operating speed and logic resources for optimal performance. Experiment results on slice data of real SAR images show that the system's processing speed can reach two frames per second, utilizing 69633 LUTs, 255 RAMs, and 296 DSPs.

The segmentation effect of style transfer on fetal head ultrasound image: a study of multi-source data.

The generalization ability of the fetal head segmentation method is reduced due to the data obtained by different machines, settings, and operations. To keep the generalization ability, we proposed a Fourier domain adaptation (FDA) method based on amplitude and phase to achieve better multi-source ultrasound data segmentation performance. Given the source/target image, the Fourier domain information was first obtained using fast Fourier transform. Secondly, the target information was mapped to the source Fourier domain through the phase adjustment parameter α and the amplitude adjustment parameter ß. Thirdly, the target image and the preprocessed source image obtained through the inverse discrete Fourier transform were used as the input of the segmentation network. Finally, the dice loss was computed to adjust α and ß. In the existing transform methods, the proposed method achieved the best performance. The adaptive-FDA method provides a solution for the automatic preprocessing of multi-source data. Experimental results show that it quantitatively improves the segmentation results and model generalization performance.