Search Results (272)

As an important part of urban ecosystems, urban green spaces play a key role in ecological environmental protection and urban spatial structure optimization. However, due to the complex morphology and high degree of fragmentation of urban green spaces, it is still challenging to effectively distinguish urban green space types from high spatial resolution images. To solve the problem, a Contextual Transformer and Squeeze Aggregated Excitation Enhanced DeepLabV3+ (CTSA-DeepLabV3+) model was proposed for urban green space classification based on Gaofen-2 (GF-2) satellite images. A Contextual Transformer (CoT) module was added to the decoder part of the model to enhance the global context modeling capability, and the SENetv2 attention mechanism was employed to improve its key feature capture ability. The experimental results showed that the overall classification accuracy of the CTSA-DeepLabV3+ model is 96.21%, and the average intersection ratio, precision, recall, and F1-score reach 89.22%, 92.56%, 90.12%, and 91.23%, respectively, which is better than DeepLabV3+, Fully Convolutional Networks (FCNs), U-Net (UNet), the Pyramid Scene Parseing Network (PSPNet), UperNet-Swin Transformer, and other mainstream models. The model exhibits higher accuracy and provides efficient references for the intelligent interpretation of urban green space with high-resolution remote sensing images. Full article

In this paper, a fully convolutional data description (FCDD) model is applied to defect detection and its concurrent visualization for industrial products and materials. The authors’ propose a MATLAB application that enables users to efficiently and in a user-friendly way design, train, and test various kinds of neural network (NN) models for defect detection. Models supported by the application include the following original designs: convolutional neural network (CNN), transfer learning-based CNN, NN-based support vector machine (SVM), convolutional autoencoder (CAE), variational autoencoder (VAE), fully convolution network (FCN) (such as U-Net), and YOLO. However, FCDD is not yet supported. This paper includes the software development of the MATLAB R2024b application, which is extended to be able to build FCDD models. In particular, a systematic threshold determination method is proposed to obtain the best performance for defect detection from FCDD models. Also, through three different kinds of defect detection experiments, the usefulness and effectiveness of FCDD models in terms of defect detection and its concurrent visualization are quantitatively and qualitatively evaluated by comparing conventional transfer learning-based CNN models. Full article

Efficient pest control in orchards is crucial for preserving crop quality and maximizing yield. A key factor in optimizing automated variable-rate spraying is accurate tree canopy size estimation, which helps reduce pesticide overuse while minimizing environmental and health risks. This study evaluates the performance of two advanced convolutional neural networks, PP-LiteSeg and fully convolutional networks (FCNs), for segmenting tree canopies of varying sizes—small, medium, and large—using short-term dense-connection networks (STDC1 and STDC2) as backbones. A dataset of 305 field-collected images was used for model training and evaluation. The results show that FCNs with STDC backbones outperform PP-LiteSeg, delivering superior semantic segmentation accuracy and background classification. The STDC1-based model excels in precision variable-rate spraying, achieving an Intersection-over-Union of up to 0.75, Recall of 0.85, and Precision of approximately 0.85. Meanwhile, the STDC2-based model demonstrates greater optimization stability and faster convergence, making it more suitable for resource-constrained environments. Notably, the STDC2-based model significantly enhances canopy-background differentiation, achieving a background classification Recall of 0.9942. In contrast, PP-LiteSeg struggles with small canopy detection, leading to reduced segmentation accuracy. These findings highlight the potential of FCNs with STDC backbones for automated apple tree canopy recognition, advancing precision agriculture and promoting sustainable pesticide application through improved variable-rate spraying strategies. Full article

Accurate segmentation of the sky region is crucial for various applications, including object detection, tracking, and recognition, as well as augmented reality (AR) and virtual reality (VR) applications. However, sky region segmentation poses significant challenges due to complex backgrounds, varying lighting conditions, and the absence of clear edges and textures. In this paper, we present a new hybrid fast segmentation technique for the sky region that learns from object components to achieve rapid and effective segmentation while preserving precise details of the sky region. We employ Convolutional Neural Networks (CNNs) to guide the active contour and extract regions of interest. Our algorithm is implemented by leveraging three types of CNNs, namely DeepLabV3+, Fully Convolutional Network (FCN), and SegNet. Additionally, we utilize a local image fitting level-set function to characterize the region-based active contour model. Finally, the Lattice Boltzmann approach is employed to achieve rapid convergence of the level-set function. This forms a deep Lattice Boltzmann Level-Set (deep LBLS) segmentation approach that exploits deep CNN, the level-set method (LS), and the lattice Boltzmann method (LBM) for sky region separation. The performance of the proposed method is evaluated on the CamVid dataset, which contains images with a wide range of object variations due to factors such as illumination changes, shadow presence, occlusion, scale differences, and cluttered backgrounds. Experiments conducted on this dataset yield promising results in terms of computation time and the robustness of segmentation when compared to state-of-the-art methods. Our deep LBLS approach demonstrates better performance, with an improvement in mean recall value reaching up to 14.45%. Full article

Applying deep neural networks (DNNs) to broadband seismic wave impedance inversion is challenging, especially in generalizing from synthetic to field data, which limits the exploitation of their nonlinear mapping capabilities. While many research studies are about advanced and enhanced architectures of DNNs, this article explores how variations in input data affect DNNs and consequently enhance their generalizability and inversion performance. This study introduces a novel data pre-processing strategy based on histogram equalization and an iterative testing strategy. By employing a U-Net architecture within a fully convolutional neural network (FCN) exclusively trained on synthetic and monochrome data, including post-stack profile, and 1D linear background impedance profiles, we successfully achieve broadband impedance inversion for both new synthetic data and marine seismic data by integrating imaging profiles with background impedance profiles. Notably, the proposed method is applied to reverse time migration (RTM) data from the Ceduna sub-basin, located in offshore southern Australia, significantly expanding the wavenumber bandwidth of the available data. This demonstrates its generalizability and improved inversion performance. Our findings offer new insights into the challenges of seismic data fusion and promote the utilization of deep neural networks for practical seismic inversion and outcomes improvement. Full article

Lamb-wave-based structural health monitoring is widely employed to detect and localize damage in composite plates; however, interpreting Lamb wave signals remains challenging due to their dispersive characteristics. Although convolutional neural networks (CNNs) demonstrate a significant capability for pattern recognition within these signals relative to other machine learning models, CNNs frequently encounter difficulties in capturing all the underlying patterns when the damage severity varies. To address this issue, we propose a multiscale, one-dimensional convolutional neural network (MS-1D-CNN) to assess the damage severity and localize damage in laminated plates. The MS-1D-CNN is capable of learning both low- and high-level features, enabling it to distinguish between minor and severe damage. The dataset was obtained experimentally via a sparse array of four lead zirconate titanates, with signals from twelve paths fused and downsampled before being input into the model. The efficiency of the model was evaluated using accuracy, precision, recall, and F1-score metrics for severity identification, along with the mean squared error, mean absolute error, and $R^2$ for damage localization. The experimental results indicated that the proposed MS-1D-CNN outperformed support vector machine and artificial neural network models, achieving higher accuracy in both identifying damage severity and localizing damage with minimal error. Full article

Communication radiation source individual identification technology is an essential technique in electronic reconnaissance and a crucial link in electronic warfare support measures. Nevertheless, when the sample set encounters complex circumstances, such as class imbalance or a small sample, the classification network model, driven by big data, disrupts the symmetry between the recognition effect and the quantity of the datasets, leading to suboptimal recognition performance. Thus, it is requisite to optimize the existing models and algorithms to better propose more representative fingerprint features. This paper references the speech signal recognition model multivariate long short-term memory–fully convolutional network (MLSTM-FCN), and ameliorates the recognition algorithm and training strategy for the two scenarios of class imbalance and a small sample. It puts forward a communication radiation source individual identification method based on MLSTM-FCN incremental random feature concatenation and a communication radiation source individual identification method based on meta-learning. Proceeding from improving the class imbalance issue among features and small-sample learning, the experimental results under various signal-to-noise ratios demonstrate that the proposed methods have superior recognition effects and higher accuracy. Full article

With the rapid development of the internet, network security threats are becoming increasingly complex and diverse, making traditional intrusion detection systems (IDSs) inadequate for handling the growing variety of sophisticated attacks. In particular, traditional methods based on rule matching and manual feature extraction demonstrate significant limitations in dealing with small samples and unknown attacks. This paper proposes an intrusion detection system based on multi-level feature extraction and inductive learning (MFEI-IDS) to address these challenges. The model innovatively integrates Fully Convolutional Networks (FCNs) with the Transformer architecture (FCN–Transformer) for feature extraction and utilizes an inductive learning component for efficient classification. The FCN–Transformer Encoder extracts multi-level features from raw network traffic, capturing local spatial patterns and global temporal dependencies, significantly enhancing the representation of network traffic while reducing reliance on manual feature engineering. The inductive learning module employs a dynamic routing mechanism to map sample feature vectors into robust class vector representations, achieving superior generalization when detecting unseen attack types. Compared to existing FCN–Transformer models, MFEI-IDS incorporates inductive learning to handle data imbalance and small-sample scenarios. Experiments on ISCX 2012 and CIC-IDS 2017 datasets show that MFEI-IDS outperforms mainstream IDS methods in accuracy, precision, recall, and F1-score, excelling in cross-dataset validation and demonstrating strong generalization capabilities. These results validate the practical potential of MFEI-IDS in small-sample learning, unknown attack detection, and dynamic network environments. Full article

A type of aquaculture pond called a dike-pond system is distributed in the low-lying river delta of China’s eastern coast. Along with the swift growth of the coastal economy, the water surfaces of the dike-pond system (WDPS) play a major role attributed to pond aquaculture yielding more profits than dike agriculture. This study aims to explore the performance of deep learning methods for extracting WDPS from high spatial resolution remote sensing images. We developed three fully convolutional network (FCN) models: SegNet, UNet, and UNet++, which are compared with two traditional methods in the same testing regions from the Guangdong–Hong Kong–Macao Greater Bay Area. The extraction results of the five methods are evaluated in three parts. The first part is a general comparison that shows the biggest advantage of the FCN models over the traditional methods is the P-score, with an average lead of 13%, but the R-score is not ideal. Our analysis reveals that the low R-score problem is due to the omission of the outer ring of WDPS rather than the omission of the quantity of WDPS. We also analyzed the reasons behind it and provided potential solutions. The second part is extraction error, which demonstrates the extraction results of the FCN models have few connected, jagged, or perforated WDPS, which is beneficial for assessing fishery production, pattern changes, ecological value, and other applications of WDPS. The extracted WDPS by the FCN models are visually close to the ground truth, which is one of the most significant improvements over the traditional methods. The third part is special scenarios, including various shape types, intricate spatial configurations, and multiple pond conditions. WDPS with irregular shapes or juxtaposed with other land types increases the difficulty of extraction, but the FCN models still achieve P-scores above 0.95 in the first two scenarios, while WDPS in multiple pond conditions causes a sharp drop in the indicators of all the methods, which requires further improvement to solve it. We integrated the performances of the methods to provide recommendations for their use. This study offers valuable insights for enhancing deep learning methods and leveraging extraction results in practical applications. Full article

As the global population ages, the friendliness of urban spaces towards seniors becomes increasingly crucial. This research primarily investigates the environmental factors that influence the safety perception of elderly people in living street spaces. Taking Dingzigu Street in Tianjin, China, as an example, by employing deep learning fully convolutional network (FCN-8s) technology and the semantic segmentation method based on computer vision, the objective measurement data of street environmental elements are acquired. Meanwhile, the subjective safety perception evaluation data of elderly people are obtained through SD semantic analysis combined with the Likert scale. Utilizing Pearson correlation analysis and multiple linear regression analysis, the study comprehensively examines the impact of the physical environment characteristics of living street spaces on the spatial safety perception of seniors. The results indicate that, among the objective environmental indicators, ① the street greening rate is positively correlated with the spatial sense of security of seniors; ② there is a negative correlation between sky openness and interface enclosure; and ③ the overall safety perception of seniors regarding street space is significantly influenced by the spatial sense of security, the sense of security during walking behavior, and the security perception in visual recognition. This research not only uncovers the impact mechanism of the street environment on the safety perception of seniors, but also offers valuable references for the age-friendly design of urban spaces. Full article

The egg has natural barriers that prevent microbiological contamination and promote food safety. The use of non-destructive methods to obtain morphometric measurements of chicken eggs has the potential to replace traditional invasive techniques, offering greater efficiency and accuracy. This paper aims to demonstrate that estimates derived from non-invasive approaches, such as 3D computed tomography (CT) image analysis, can be comparable to conventional destructive methods. To achieve this goal, two widely recognized deep learning architectures, U-Net 3D and Fully Convolutional Networks (FCN) 3D, were modeled to segment and analyze 3D CT images of chicken eggs. A dataset of real CT images was created and labeled, allowing the extraction of important morphometric measurements, including height, width, shell thickness, and volume. The models achieved an accuracy of up to 98.69%, demonstrating their effectiveness compared to results from manual measurements. These findings highlight the potential of CT image analysis, combined with deep learning, as a non-invasive alternative in industrial and research settings. This approach not only minimizes the need for invasive procedures but also offers a scalable and reliable method for egg quality assessment. Full article

Women face disadvantages in urban public spaces due to their physiological characteristics. However, limited attention has been given to assessing safety perceptions from a female perspective and identifying the factors that influence these perceptions. Despite advancements in machine learning (ML) techniques, efficiently and accurately quantifying safety perceptions remains a challenge. This study, using Wuhan as a case study, proposes a method for ranking street safety perceptions for women by combining RankNet with Gist features. Fully Convolutional Network-8s (FCN-8s) was employed to extract built environment features, while Ordinary Least Squares (OLS) regression and Geographically Weighted Regression (GWR) were used to explore the relationship between these features and women’s safety perceptions. The results reveal the following key findings: (1) The safety perception rankings in Wuhan align with its multi-center urban pattern, with significant differences observed in the central area. (2) Built environment features significantly influence women’s safety perceptions, with the Sky View Factor, Green View Index, and Roadway Visibility identified as the most impactful factors. The Sky View Factor has a positive effect on safety perceptions, whereas the other factors exhibit negative effects. (3) The influence of built environment features on safety perceptions varies spatially, allowing the study area to be classified into three types: sky- and road-dominant, building-dominant, and greenery-dominant regions. Finally, this study proposes targeted strategies for creating safer and more female-friendly urban public spaces. Full article

Augmented reality applications involving human interaction with virtual objects often rely on segmentation-based hand detection techniques. Semantic segmentation can then be enhanced with instance-specific information to model complex interactions between objects, but extracting such information typically increases the computational load significantly. This study proposes a training strategy that enables conventional semantic segmentation networks to preserve some instance information during inference. This is accomplished by introducing pixel weight maps into the loss calculation, increasing the importance of boundary pixels between instances. We compare two common fully convolutional network (FCN) architectures, U-Net and ResNet, and fine-tune the fittest to improve segmentation results. Although the resulting model does not reach state-of-the-art segmentation performance on the EgoHands dataset, it preserves some instance information with no computational overhead. As expected, degraded segmentations are a necessary trade-off to preserve boundaries when instances are close together. This strategy allows approximating instance segmentation in real-time using non-specialized hardware, obtaining a unique blob for an instance with an intersection over union greater than 50% in 79% of the instances in our test set. A simple FCN, typically used for semantic segmentation, has shown promising instance segmentation results by introducing per-pixel weight maps during training for light-weight applications. Full article

In order to improve the prediction accuracy of short-term power loads in a power system, this paper proposes a short-term load prediction method (POA-DLSTMs-FCN) based on a combination of multi-layer lost long short-term memory (DLSTM) neural networks, fully convolutional neural networks (FCNs) and the pelican optimization algorithm (POA). This method firstly uses DLSTMs to extract the time-series features of the load data, which can effectively capture the dynamic changes in the time series; subsequently, it combines the convolution operation of FCNs to obtain high-resolution information between the load data and the features, which enhances the expressive ability of the model. Through a parallel structure, DLSTMs and FCNs can jointly optimize the information extraction and then construct a more accurate load forecasting model. In addition, the learning rate, the number of hidden neurons and the deactivation probability of the Dropout layer in DLSTMs are optimized by the POA to further enhance the performance of the model. The experimental results show that the proposed optimization method has significant advantages over traditional DLSTMs and FCN-LSTM models in terms of prediction accuracy and stability. Full article

Packaging design plays a critical role in brand recognition and cultural dissemination, yet the traditional design process is time-consuming and dependent on the designer’s technical skills, making it difficult to quickly respond to market changes and consumer demands. In recent years, advancements in machine learning, particularly in the field of natural language processing (NLP), have paved the way for novel methods in other areas, such as image processing and packaging design. This study draws inspiration from advanced NLP techniques and proposes an improved fully convolutional network (FCN) model for image semantic segmentation, which is applied to packaging design. The model integrates superpixel technology, multi-branch networks, dual-attention mechanisms, and edge knowledge distillation in a manner analogous to the approach taken by NLP models in the context of semantic segmentation and context understanding. The experimental results showed that the model achieved significant improvements in accuracy, inference efficiency, and memory usage, with an average accuracy of 96.84% and a false-alarm rate of only 2.78%. Compared to traditional methods, the proposed model achieved over 96% accuracy across 50 packaging design images, with an average segmentation error rate of only 1.42%. By incorporating machine learning techniques from NLP into image processing, this study enhances the overall quality and efficiency of packaging design and provides new directions for the application of advanced technologies across different fields. Full article