In recent years, significant progress has been witnessed in the field of deep learning-based object detection. As a subtask detection, traffic sign detection great potential for development. However, existing methods real-world scenes are plagued by issues such as omission small objects and low accuracies. To address these issues, model named YOLOv7-Traffic Sign (YOLOv7-TS) is proposed based on sub-pixel convolution feature fusion. Firstly, up-sampling capability integrating channel...

Much attention has been paid to construct an applicable knowledge measure or uncertainty for Atanassov's intuitionistic fuzzy set (AIFS). However, many of these measures were developed from entropy, which cannot really reflect the amount associated with AIFS well. Some constructed based on distinction between and its complementary set, may lead information loss in decision making. In this paper, is quantified by calculating distance maximum uncertainty. Axiomatic properties definition are...

Abstract Excellent performance has been demonstrated in implementing challenging agricultural production processes using modern information technology, especially the use of artificial intelligence methods to improve environments. However, most existing work uses visual train models that extract image features organisms analyze their behavior, and it may not be truly intelligent. Because vocal animals transmit through grunts, obtained directly from grunts pigs is more useful understand...

Extracting water bodies from remote sensing images is important in many fields, such as resources information acquisition and analysis. Conventional methods of body extraction enhance the differences between other interfering to improve accuracy boundary extraction. Multiple must be used alternately extract boundaries more accurately. Water combined with neural networks struggle fine while ensuring an overall effect. In this study, false color processing a generative adversarial network...

Antenna effects on a monostatic multiple-input-multiple-output (MIMO) Radar for direction estimation are studied by analyzing the Cramèr-Rao lower bound (CRLB). The CRLB is derived multi-band MIMO system, and in form that incorporates characteristics of practical antenna array. Two different uniform linear arrays, one narrowband another wideband, investigated exploring CRLB. performance real array compared to an ideal composed isotropic elements. It found out mutual coupling between elements...

Imbalanced datasets greatly affect the analysis capability of intrusion detection models, biasing their classification results toward normal behavior and leading to high false-positive false-negative rates. To alleviate impact class imbalance on accuracy network models improve effectiveness, this paper proposes a method based feature selection-conditional Wasserstein generative adversarial (FCWGAN) bidirectional long short-term memory (BiLSTM). The uses XGBoost algorithm with Spearman's...

Abstract Currently, many real-time semantic segmentation networks aim for heightened accuracy, inevitably leading to increased computational complexity and reduced inference speed. Therefore, striking a balance between accuracy speed has emerged as crucial concern in this domain. To address these challenges, study proposes dual-branch fusion network with multiscale atrous pyramid pooling aggregate contextual features (MAFNet). The first key component, the semantics guide spatial-details...

Remote sensing image object detection (RSIOD) aims to identify and locate specific objects within satellite or aerial imagery. However, there is a scarcity of labeled data in current RSIOD datasets, which significantly limits the performance algorithms. Although existing techniques, e.g., augmentation semi-supervised learning, can mitigate this issue some extent, they are heavily dependent on high-quality perform worse rare classes. To address issue, paper proposes layout-controllable...

The sound of the pig is one its important signs, which can reflect various states such as hunger, pain or emotional state, and directly indicates growth health status pig. Existing speech recognition methods usually start with spectral features. use spectrograms to achieve classification different sounds, while working well, may not be best approach for solving tasks single-dimensional feature input. Based on above assumptions, in order more accurately grasp situation pigs take timely...

Currently, malware shows an explosive growth trend. Demand for classifying is also increasing. The problem the low accuracy of both detection and classification. From static features malicious families, a new deep learning method TCN-BiGRU was proposed in this study, which combined temporal convolutional network (TCN) bidirectional gated recurrent unit (BiGRU). First, we extracted assembly code sequences byte sequences. Second, shortened opcode by TCN to explore data then used BiGRU capture...

Abstract Recently, owing to the requirements of inference speed, most real-time semantic segmentation networks often have shallow network depth, which limits receptive field size model, leading limited acquisition information and resulting in intraclass inconsistency ultimately a decrease accuracy. Additionally, depth also restricts feature extraction capability network, reducing its robustness ability adapt complex scenes. To address these issues, bilateral with rich extractor (RSE) for...

The detection of traversable regions on staircases and the physical modeling constitutes pivotal aspects mobility legged robots. This paper presents an onboard framework tailored to attributes by point cloud data. To mitigate influence illumination variations overfitting due dataset diversity, a series data augmentations are introduced enhance training fundamental network. A curvature suppression cross-entropy(CSCE) loss is proposed reduce ambiguity prediction boundary between...

Abstract Radar signal sorting is a vital component of electronic warfare reconnaissance, serving as the basis for identifying source radar signals. However, traditional methods are increasingly inadequate and computationally complex in modern electromagnetic environments. To address this issue, paper presents novel machine-learning-based approach sorting. Our method utilizes SemHybridNet, Semantically Enhanced Hybrid CNN-Transformer Network, classification semantic information...

Computer vision (CV) techniques have been widely studied and applied in the shipping industry maritime research. The existing literature has primarily focused on enhancing image recognition accuracy precision for water surface targets by refining CV models themselves. This paper introduces innovative methods to further improve of detection using models, including ensemble learning integrating domain knowledge. Additionally, we present a novel application domain, expanding research...

In recent years, the presence of malware has been growing exponentially, resulting in enormous demand for efficient classification methods. However, existing machine learning-based classifiers have high false positive rates and cannot effectively classify variants, packers, obfuscation. To address this shortcoming, paper proposes an deep method named AIFS-IDL (Atanassov Intuitionistic Fuzzy Sets-Integrated Deep Learning), which uses static features to malware. The proposed first extracts six...

This paper analyzes the dynamic and incomplete information characteristics of network offensive defensive confrontation. In this paper, a signal game is used as framework, defender initiator. The behavior model that induces to interfere with attack construct moving target defense game. Considering unavoidable misdetection defects detection system itself, method for quantifying benefits strategy proposed, refined Bayesian equilibrium algorithm prior belief correction are given. optimal...

Aiming at the problems of unclear boundaries and low segmentation accuracy in general real-time semantic networks, a network based on improved BiSeNet V1 was proposed. Based BiSeNetV1 network, spatial enhancement module (SRM) is introduced into path to enhance information improve detection ability target small targets. At same time, when context feature are fused, Feature Aggregation Module (FAM) proposed solve difference representation between two paths fusion efficiency. We experiment...

The existing supervised learning methods can only use labelled samples to train the classifier, which is difficult and costly obtain labels. To solve problem enhance effectiveness of intrusion detection models, a semi-supervised method proposed in this study terms based on Fuzzy-Long Short-Term Memory (Fuzzy-LSTM). model uses long short-term memory generate labels for unlabeled samples, while classifying fuzzy entropy. low entropy from them were merged into original training set, classifier...