Scale Variations Research Articles

Robust Multimodal Image Matching Based on Radiation Invariant Phase Correlation

Abstract. Due to the influence of nonlinear radiation distortion and geometric deformation, achieving multimodal image matching remains a challenging task. To address these issues, this paper proposes a method called radiation invariant phase correlation (RIPC) to simultaneously estimate the rotation, scale, and displacement changes of multimodal image pairs. Firstly, based on the local structure characteristics of the image itself, we harness the nonlinear invariance of kernel canonical correlation analysis to devise the multimodal local self-correlation (MLSC) descriptor. This descriptor is resilient to nonlinear radiative differences, as well as local rotation and scale variations. Subsequently, we incorporate the log-polar coordinate transformation to capture the overall rotation and scale changes in the image, enabling independent representation of these factors on the Cartesian coordinate system. Finally, drawing upon the continuity of displacement estimation, as well as rotation and scale estimation, we construct a five-dimensional descriptor tailored for phase correlation. Extensive experiments conducted on five open-source datasets demonstrate that our proposed method surpasses state-of-the-art (SOTA) techniques in matching performance. Furthermore, our RIPC method achieves matching accuracy within 2-pixel threshold, which underscores its effectiveness in multimodal remote sensing image matching.

Multi‐scale feature extraction for energy‐efficient object detection in remote sensing images

AbstractObject detection in remote sensing images aims to interpret images to obtain information on the category and location of potential targets, which is of great importance in traffic detection, marine supervision, and space reconnaissance. However, the complex backgrounds and large scale variations in remote sensing images present significant challenges. Traditional methods relied mainly on image filtering or feature descriptor methods to extract features, resulting in underperformance. Deep learning methods, especially one‐stage detectors, for example, the Real‐Time Object Detector (RTMDet) offers advanced solutions with efficient network architectures. Nevertheless, difficulty in feature extraction from complex backgrounds and target localisation in scale variations images limits detection accuracy. In this paper, an improved detector based on RTMDet, called the Multi‐Scale Feature Extraction‐assist RTMDet (MRTMDet), is proposed which address limitations through enhancement feature extraction and fusion networks. At the core of MRTMDet is a new backbone network MobileViT++ and a feature fusion network SFC‐FPN, which enhances the model's ability to capture global and multi‐scale features by carefully designing a hybrid feature processing unit of CNN and a transformer based on vision transformer (ViT) and poly‐scale convolution (PSConv), respectively. The experiment in DIOR‐R demonstrated that MRTMDet achieves competitive performance of 62.2% mAP, balancing precision with a lightweight design.

Enhanced Hybrid U-Net Framework for Sophisticated Building Automation Extraction Utilizing Decay Matrix

Automatically extracting buildings from remote sensing imagery using deep learning techniques has become essential for various real-world applications. However, mainstream methods often encounter difficulties in accurately extracting and reconstructing fine-grained features due to the heterogeneity and scale variations in building appearances. To address these challenges, we propose LDFormer, an advanced building segmentation model based on linear decay. LDFormer introduces a multi-scale detail fusion bridge (MDFB), which dynamically integrates shallow features to enhance the representation of local details and capture fine-grained local features effectively. To improve global feature extraction, the model incorporates linear decay self-attention (LDSA) and depthwise large separable kernel multi-layer perceptron (DWLSK-MLP) optimizations in the decoder. Specifically, LDSA employs a linear decay matrix within the self-attention mechanism to address long-distance dependency issues, while DWLSK-MLP utilizes step-wise convolutions to achieve a large receptive field. The proposed method has been evaluated on the Massachusetts, Inria, and WHU building datasets, achieving IoU scores of 76.10%, 82.87%, and 91.86%, respectively. LDFormer demonstrates superior performance compared to existing state-of-the-art methods in building segmentation tasks, showcasing its significant potential for building automation extraction.

GCS-YOLOv8: A Lightweight Face Extractor to Assist Deepfake Detection

To address the issues of target feature blurring and increased false detections caused by high compression rates in deepfake videos, as well as the high computational resource requirements of existing face extractors, we propose a lightweight face extractor to assist deepfake detection, GCS-YOLOv8. Firstly, we employ the HGStem module for initial downsampling to address the issue of false detections of small non-face objects in deepfake videos, thereby improving detection accuracy. Secondly, we introduce the C2f-GDConv module to mitigate the low-FLOPs pitfall while reducing the model’s parameters, thereby lightening the network. Additionally, we add a new P6 large target detection layer to expand the receptive field and capture multi-scale features, solving the problem of detecting large-scale faces in low-compression deepfake videos. We also design a cross-scale feature fusion module called CCFG (CNN-based Cross-Scale Feature Fusion with GDConv), which integrates features from different scales to enhance the model’s adaptability to scale variations while reducing network parameters, addressing the high computational resource requirements of traditional face extractors. Furthermore, we improve the detection head by utilizing group normalization and shared convolution, simplifying the process of face detection while maintaining detection performance. The training dataset was also refined by removing low-accuracy and low-resolution labels, which reduced the false detection rate. Experimental results demonstrate that, compared to YOLOv8, this face extractor achieves the AP of 0.942, 0.927, and 0.812 on the WiderFace dataset’s Easy, Medium, and Hard subsets, representing improvements of 1.1%, 1.3%, and 3.7% respectively. The model’s parameters and FLOPs are only 1.68 MB and 3.5 G, reflecting reductions of 44.2% and 56.8%, making it more effective and lightweight in extracting faces from deepfake videos.

Surface defect inspection of industrial products with object detection deep networks: a systematic review

One of the focal points in industrial product defect detection lies in the utilization of deep learning-based object detection algorithms. With the continuous introduction of these algorithms and their refined models, notable achievements have been attained. However, challenges persist in industrial settings, such as substantial variations in defect scales, the delicate balance between accuracy and speed, and the detection of small objects. Various methods have been proposed to address these challenges and propel the advancement of defect detection. To comprehensively review the latest developments in deep learning-based industrial product defect detection algorithms and foster further progress, this paper encompasses typical datasets and evaluation metrics used in industrial product defect detection, traces the development history of supervised one-stage and two-stage object detection algorithm-based and unsupervised algorithm-based industrial defect detection methods, discusses major challenges, and outlines future directions. It highlights the potential for further improving the accuracy, speed, and reliability of defect detection systems in industrial applications.

Plasma C-peptide mammographic features and risk of breast cancer

Our study in the Nurses’ Health Study (NHS) and NHS2, a nested case-control study with 1260 cases and 2221 controls, investigated the association between C-peptide levels, mammographic density (MD) parameters, V (a measure of gray scale variation), and breast cancer (BC) risk. We also examined how C-peptide and BC risk vary across quartiles of mammographic features. Linear and logistic regressions were used to study the associations between C-peptide and MD parameters, and breast cancer. C-peptide was inversely associated with percent MD and positively with non-dense area, but no associations were found with dense area and V measure. C-peptide was associated with an increased risk of invasive BC risk (top vs. bottom quartile, odds ratio = 1.46, 95% CI: 1.12–1.91). No multiplicative interactions were found between C-peptide, MD parameters, and BC risk. Our results suggest a positive association between C-peptide and BC risk, and MD parameters do not seem to modify this association.

Insights into the plankton community seasonal variations in a finer scale of the Bohai Sea: biodiversity, trophic linkage, and biotic-abiotic interplay

Plankton play an indispensable role in the biogeochemical processes of marine ecosystem. However, unraveling the intricate interactions among biodiversity, trophic linkages, and biotic-abiotic interplay between phytoplankton-zooplankton remains a significant challenge. Here, we conducted field studies in the neritic area of the Bohai Sea during autumn 2023 and spring 2024 to explore seasonal variations of both phytoplankton and zooplankton through microscope. Our analysis revealed a sharp decline in trophic interactions across phytoplankton and zooplankton, with an abundance ratio in autumn 2023 being 5.5 times higher than in spring 2024. Additionally, dominant plankton species (Y ≥ 0.02) exhibited obvious differences between the two seasons, with higher species diversity observed in autumn. Moreover, each dominant zooplankton species had distinct preferred food items in both seasons, with Rhizosolenia setigera being favored by Noctiluca scintillans and Acartia pacifica. Furthermore, a multivariate biota-environment analysis indicated that each dominant plankton species had unique correlation with specific environmental parameters, highlighting how plankton can fully exploit external environmental conditions to survive in seasonal variations. Ultimately, our findings emphasize significant seasonal dynamics and provide a solid foundation for assessing the potential impacts of environmental changes on plankton in coastal marine realm.

SODD-YOLOv8: an insulator defect detection algorithm based on feature enhancement and variable row convolution

Abstract To address the detection challenges in defective images of insulators in transmission lines, including tiny object size, significant scale variations, a wide variety of defects, and complex background interference. In this study, an improved insulator defect detection algorithm is proposed, based on the YOLOv8s framework and combining feature enhancement and deformable convolution techniques. Firstly, to address the image feature distortion problem caused by aerial photography, a deformable convolutional feature extraction module (DCFEM) is introduced, which is designed to enhance the model’s ability to adapt to the local geometric deformation, so as to effectively recover the distorted feature information in the image. Moreover, to enhance the detection ability of the model for small objects, a small object feature enhancement module is designed, which adopts an efficient multi-scale attention mechanism, and aims to enhance the feature extraction ability of small objects, improve the sensitivity to small-size defects, and improve the detection accuracy. Eventually, to optimize the computational efficiency of the model, the average pooling-sparse convolution-batch normalization (BN) module is proposed. This module combines average pooling, sparse convolution and BN techniques to achieve a lightweight model while maintaining a high level of feature extraction capability. Experimental results on the China power line insulator dataset show that the improved model achieves a 4.3 percentage point improvement in the mAP metric compared to YOLOv8s, and the number of parameters in the model is reduced by 10%. The proposed scheme not only improves the accuracy and efficiency of defect detection, but also reduces the demand for computational resources, thus providing a more reliable and efficient solution for insulator defect detection in practical applications.

Semantic Segmentation of Satellite Images for Landslide Detection Using Foreground-Aware and Multi-Scale Convolutional Attention Mechanism.

Advancements in satellite and aerial imagery technology have made it easier to obtain high-resolution remote sensing images, leading to widespread research and applications in various fields. Remote sensing image semantic segmentation is a crucial task that provides semantic and localization information for target objects. In addition to the large-scale variation issues common in most semantic segmentation datasets, aerial images present unique challenges, including high background complexity and imbalanced foreground-background ratios. However, general semantic segmentation methods primarily address scale variations in natural scenes and often neglect the specific challenges in remote sensing images, such as inadequate foreground modeling. In this paper, we present a foreground-aware remote sensing semantic segmentation model. The model introduces a multi-scale convolutional attention mechanism and utilizes a feature pyramid network architecture to extract multi-scale features, addressing the multi-scale problem. Additionally, we introduce a Foreground-Scene Relation Module to mitigate false alarms. The model enhances the foreground features by modeling the relationship between the foreground and the scene. In the loss function, a Soft Focal Loss is employed to focus on foreground samples during training, alleviating the foreground-background imbalance issue. Experimental results indicate that our proposed method outperforms current state-of-the-art general semantic segmentation methods and transformer-based methods on the LS dataset benchmark.

MCH-PAN: gastrointestinal polyp detection model integrating multi-scale feature information.

The rise of object detection models has brought new breakthroughs to the development of clinical decision support systems. However, in the field of gastrointestinal polyp detection, there are still challenges such as uncertainty in polyp identification and inadequate coping with polyp scale variations. To address these challenges, this paper proposes a novel gastrointestinal polyp object detection model. The model can automatically identify polyp regions in gastrointestinal images and accurately label them. In terms of design, the model integrates multi-channel information to enhance the ability and robustness of channel feature expression, thus better coping with the complexity of polyp structures. At the same time, a hierarchical structure is constructed in the model to enhance the model's adaptability to multi-scale targets, effectively addressing the problem of large-scale variations in polyps. Furthermore, a channel attention mechanism is designed in the model to improve the accuracy of target positioning and reduce uncertainty in diagnosis. By integrating these strategies, the proposed gastrointestinal polyp object detection model can achieve accurate polyp detection, providing clinicians with reliable and valuable references. Experimental results show that the model exhibits superior performance in gastrointestinal polyp detection, which helps improve the diagnostic level of digestive system diseases and provides useful references for related research fields.

Characteristic analysis of India-Burma trough events and its impact on winter precipitation in South and East Asia: Based on objective identification

Abstract During the boreal winter, the India-Burma trough (IBT), a shortwave trough system primarily positioned over the northern Bay of Bengal, exerts a significant synoptic scale variation and impact on precipitation in South and East Asia. This study utilizes the 6-hourly ERA5 dataset to objectively identify and track 714 IBT events from 1981 to 2019. On average, IBT occurred about 54.7 days per year, with an average of 18.3 events annually and lasting around 2.5 days. IBT events are classified into two types: local and eastward-moving. Local IBTs manifest in the middle and lower troposphere with a vertical temperature structure of warm-over-cold, whereas the signals of the eastward-moving IBTs extend from the lower to the upper troposphere, exhibiting cold anomalies vertically. The impacts on winter precipitation in South and East Asia differ significantly between these two IBT types. Local IBTs are primarily linked with the eastward propagation of wave disturbances along the subtropical westerly jet, while eastward-moving IBTs are associated with robust disturbance sources from mid-high latitudes and jointly modulated by the consequent eastward propagation of Rossby wave trains along northern and southern branches of the westerly jet streams. Precipitation anomalies during local IBTs are typically positive (negative) in the Indochina Peninsula and southwestern China (Indian Peninsula), whereas eastward-moving IBTs correlate with more intense and widespread precipitation in South and East Asia, with a pronounced band of anomalies from the Indochina Peninsula to southern China.

Mutually-Guided Hierarchical Multi-Modal Feature Learning for Referring Image Segmentation

Referring image segmentation aims to locate and segment the target region based on a given textual expression query. The primary challenge is to understand semantics from visual and textual modalities and achieve alignment and matching. Prior works have attempted to address this challenge by leveraging separately pretrained unimodal models to extract global visual and textual features, and perform straightforwardly fusion to establish cross-modal semantic associations. However, these methods often concentrate solely on the global semantics, disregarding the hierarchical semantics of expression and image, and struggling with complex and open real scenarios, thus failing to capture critical cross-modal information. To address these limitations, this paper introduces an innovative mutually-guided hierarchical multi-modal feature learning scheme. By leveraging the guidance of global visual features, the model mines hierarchical text features from different stages of the text encoder. Simultaneously, the guidance of global textual features is leveraged to aggregate multi-scale visual features. This mutually guided hierarchical feature learning effectively addresses the semantically inaccurate cause by free-form text and naturally occurring scale variations. Furthermore, a Segment Detail Refinement (SDR) module is designed to enhance the model’s spatial detail awareness through attention mapping of low-level visual features and cross-modal features. To evaluate the effectiveness of the proposed approach, extensive experiments are conducted on three widely used referring image object segmentation datasets. The results demonstrate the superiority of the presented method in accurately locating and segmenting objects in images.

E2TNet: Efficient enhancement Transformer network for hyperspectral image classification

Recently, Convolutional Transformer-based models have become popular in hyperspectral image (HSI) classification tasks and gained competitive classification performance. However, some Convolutional Transformer-based models fail to effectively mine the global correlations of coarse-grained and fine-grained features, which is adverse to recognizing the refined scale variation information of land-cover. The combination of convolution operations and multihead self-attention mechanisms also increases the computational cost, leading to low classification efficiency. In addition, shallow spectral–spatial features are directly input into the encoder, which inevitably incurs redundant spectral information. Therefore, this paper proposes an efficient enhancement Transformer network (E2TNet) for HSI classification. Specifically, this paper first designs a spectral–spatial feature fusion module to extract spectral and spatial features from HSI cubes and fuse them. Second, considering that redundant spectral information has a negative impact on classification performance, this paper designs a spectral–spatial feature weighted module to improve the feature representation of critical spectral information. Finally, to explore the global correlations of coarse-grained and fine-grained features and improve classification efficiency, an efficient multigranularity information fusion module is embedded in the encoder of E2TNet. The experiment is conducted on four benchmark hyperspectral datasets, and the experimental results demonstrate that the proposed E2TNet is better than several Convolutional Transformer-based classification models in terms of classification accuracy and classification efficiency.

SSN: Scale Selection Network for Multi-Scale Object Detection in Remote Sensing Images

The rapid growth of deep learning technology has made object detection in remote sensing images an important aspect of computer vision, finding applications in military surveillance, maritime rescue, and environmental monitoring. Nonetheless, the capture of remote sensing images at high altitudes causes significant scale variations, resulting in a heterogeneous range of object scales. These varying scales pose significant challenges for detection algorithms. To solve the scale variation problem, traditional detection algorithms compute multi-layer feature maps. However, this approach introduces significant computational redundancy. Inspired by the mechanism of cognitive scaling mechanisms handling multi-scale information, we propose a novel Scale Selection Network (SSN) to eliminate computational redundancy through scale attentional allocation. In particular, we have devised a lightweight Landmark Guided Scale Attention Network, which is capable of predicting potential scales in an image. The detector only needs to focus on the selected scale features, which greatly reduces the inference time. Additionally, a fast Reversible Scale Semantic Flow Preserving strategy is proposed to directly generate multi-scale feature maps for detection. Experiments demonstrate that our method facilitates the acceleration of image pyramid-based detectors by approximately 5.3 times on widely utilized remote sensing object detection benchmarks.

Lightweight algorithm based on you only look once version 5 for multiple class defect detection on wind turbine blade surfaces

A lightweight wind turbine blade surface multiple class defect detection algorithm based on You Only Look Once version 5 (YOLOv5) is proposed to address the issues of low accuracy, poor efficiency, and high power consumption associated with manual and traditional automation algorithms used for detecting surface defects on wind turbine blades. Firstly, by replacing the Concentrated-Comprehensive Convolution Block module in YOLOv5 with the GhostBottleneckv2 module, the overall parameter count of the model is significantly reduced. Additionally, advocating the inclusion of the Convolutional Block Attention Module (CBAM) into the YOLOv5 backbone network to address the issue of small targets and significant scale variations in surface defects found on wind turbine blades. The YOLOv5 feature fusion module replaces the Feature Pyramid Network (FPN) structure with the Bidirectional Feature Pyramid Network (BiFPN) structure to enhance the fusion capability of multi-scale weighted features. Finally, the optimized model files underwent structured pruning through the application of pruning strategies, retraining fine-tuning, and knowledge distillation techniques. The experimental results on the self-made dataset of surface defects on wind turbine blades indicate that the optimized model has a parameter count that is only 48.79% of the small version of the YOLOv5 model. However, its mean Average Precision (mAP) and frame rate have increased by 5.05% and 8.1 frames/s, respectively, reaching 94.56% and 43.7 frames/s. The lightweight algorithm proposed in this study meets the accuracy and real-time requirements for surface defect detection on wind turbine blades.

Revealing the Intrinsic Correlation between Cu Scales and Free Radical Chain Reactions in the Regulation of Catalytic Behaviour.

Defining the copper-based catalysts that are responsible for the catalytic behaviour of oil-paper insulation systems and implementing effective regulation are of great significance. Accelerated ageing experiments were conducted to reveal variations in copper scales and deterioration in insulation properties. As ageing progressed, TEM images demonstrated that copper species were adsorbed and aggregated on the fibre surface in the form of nanoparticles (NPs). The scale of NPs exhibited a continuous increase, from 27.06 nm to 94.19 nm. Cu(I) and Cu(II) species were identified as the active sites for inducing intense free radical reactions, which significantly reduced the activation energy, making the insulating oil more susceptible to oxidation. The role of the antioxidant di-tert-butyl-p-cresol (DBPC) in extending the insulation life was regulated by determining the optimal addition time based on variations in the interfacial tension. After the second addition of DBPC, the ageing rates of the dissipation factor, acidity, micro-water and breakdown voltage in the Cu+DBPC group decreased by 28.8%, 43.2%, 52.9% and 46.7%, respectively, compared to the Cu group. This finding not only demonstrates the crucial role of DBPC in preventing the copper-based catalyst-induced oxidation of insulating oil, but also furnishes a vital foundation for enhancing the long-term stability of transformer insulation systems.

Characterizing and predicting the partial slip subjected to variable gradients of velocity and pressure in eccentric micro-scale Taylor–Couette flows

In the context of eccentric micro-scale Taylor–Couette flow, variations in localized flow scales result in a non-uniform fluid–solid interface slip state, distinct from the typically studied uniform slip velocity distribution. This study introduces a boundary condition definition method aimed at characterizing partial slip states, complemented by a coupled iterative analysis system tailored to address this complexity. Key contributions include the development of a method for calculating the limiting shear stress, which considers local velocity gradients and pressures. Validation demonstrates that the locally derived slip state aligns more closely with Knudsen number distributions of local flow scales compared to traditional uniform slip models, and exhibits greater consistency with experimentally measured pressure distributions. Additionally, the study reveals that eccentric Taylor–Couette flow, characterized by significant variations in local flow scales and strong self-induced pressure effects, leads to complex distributions of local pressure, velocity gradients, and differences in local slip velocities. Specifically, the non-uniform distribution of local pressure gradients due to eccentricity results in partial slip occurring predominantly on the rotor in regions with positive pressure gradients, and on the stator in regions with negative pressure gradients. Furthermore, the variation in gap height exerts a greater influence on local slip compared to rotational speed and eccentricity ratio. Under certain conditions influenced by pressure gradients, the slip velocity on the rotor may exceed its tangential speed.

AVA-YOLO: image-based multiscale feature fusion enhanced perception model for snow avalanche detection

Abstract The global climate change has led to frequent occurrences of snow avalanche disasters. However, the significant variations in scale and shape during the avalanche process, and complex background imagery pose significant challenges to automated detection efforts. There is an urgent need to combine advanced deep learning technology to research automatic detection and recognition of avalanches in the field. In this paper, a novel deep learning model based on YOLOv8 improved multi-scale detection called AVA-YOLO is proposed to solve this problem. In AVA-YOLO, a key component, AKA (AKConv Combined Attention) module was designed and developed. This module combines the deformable convolutional properties of AKConv with the state-of-the-art self-attention module Exponential Moving Average, aiming to better perceive the feature map information of different shaped avalanches and to enhance the global relevance, thus improving the utilization of the information. Secondly, a new multi-scale sensing network structure was designed by increasing the number of detection heads to four and introducing the AKA module into the key positions of the network, while the association between model layers was newly designed to enhance the fusion of shallow and deep information to improve the detection accuracy. Experimental results demonstrated the effectiveness of AVA-YOLO, achieving 95.7% mAP50 and 75.6% mAP50:95 detection accuracies, as well as an F1 score of 0.92. Finally, a number of experiments were conducted to demonstrate the superior performance of the proposed model in comparison to other versions of YOLO, which will further exploit the potential of webcams as an underutilized technical capability in snow avalanche intelligence and portable monitoring.

Research on Optimization of Visual Object Tracking Algorithm Based on Deep Learning

The appearance of deep gaining knowledge has extensively advanced the sphere of visible item tracking, permitting greater robust and accurate monitoring of items throughout complex scenes. This study optimises a visual item tracking set of rules based on the Siamese region inspiration network (Siamese RPN) monitoring algorithm, aiming to beautify its efficiency and effectiveness in actual-time packages. The Siamese RPN algorithm, acknowledged for its stability among accuracy and velocity because of its architecture that mixes the Siamese network for characteristic extraction with a vicinity suggestion community for item localisation, provides a promising basis for improvement. This examination introduces numerous optimisations to the authentic Siamese RPN framework. First, we endorse an improved feature extraction model that leverages a more efficient deep neural community structure, lowering computational load while preserving excessive accuracy. 2nd, we optimise the place concept mechanism by incorporating an adaptive anchor scaling method that dynamically adjusts the scale and ratio of anchors based on the object’s scale variations, enhancing the tracking accuracy across distinctive object sizes and aspect ratios. Moreover, we introduce a unique training method that employs an aggregate of actual global and synthetically generated statistics to beautify the robustness of the monitoring algorithm towards various demanding situations, including occlusions, speedy moves, and illumination adjustments. The effectiveness of the proposed optimizations is evaluated through complete experiments on numerous benchmark datasets, consisting of OTB, VOT, and LaSOT, demonstrating extensive upgrades in tracking accuracy and speed as compared to the authentic Siamese RPN algorithm and different modern-day tracking techniques. The outcomes of this study no longer underscore the potential of optimised Siamese RPN algorithms in visible item monitoring but additionally lay the basis for future explorations into actual-time, green, and strong tracking systems. Those improvements keep great promise for a wide variety of packages, from surveillance and protection to autonomous cars and augmented truth structures, where particular and dependable item monitoring is paramount.

Lightweight UAV target detection algorithm

Abstract Aiming at the public area UAV detection task, the target is easy to occlude, the scale variation is large, and the existing detection algorithm model parameters are large. In this paper, the lightweight network model YOLOv5s-UD is designed for the low-altitude UAV target detection algorithm. In this detection model, the proposed approach employs the streamlined Shufflenetv2 architecture to enhance the core network. To optimize the UAV’s relevant feature data, the CBAM attention module is integrated during the feature amalgamation phase. Furthermore, the GSConv is implemented to upgrade conventional convolution within the neck network, ensuring a reduction in model parameters without compromising detection precision. Experimental outcomes reveal a 36.2% computational reduction compared to the YOLOv5s algorithm, with an average detection accuracy of 84.1%. This demonstrates the algorithm’s capability to efficiently detect UAV targets while minimizing computational loads, thereby exhibiting significant practical significance.