Abstract Crack information provides important evidence of structural degradation and safety in civil structures. Existing inspection methods are inefficient difficult to rapidly deploy. A real‐time crack method is proposed this study address difficulty. Within method, a wall‐climbing unmanned aerial system (UAS) developed acquire detailed images without distortion, then wireless data transmission applied fulfill detection requirements, allowing smartphones receive video taken from the UAS....

Abstract Cables and hangers are critical components of long‐span bridges, tension forces them needed to be accurately measured for ensuring the safety bridges. Traditionally, cable by attached accelerometers or elastomagnetic (EM) sensors, however, applying these sensors into engineering practice time‐consuming, labor‐intensive, highly dangerous. To address problems, an unmanned aerial vehicle (UAV)‐based noncontact force estimation method with computer vision technologies was proposed in...

Bolted connections are essential components that require regular inspection to ensure bridge safety. Existing methods mainly rely on traditional artificial vision-based inspection, which is inefficient due the many bolts of bridges. A method using deep learning and unmanned aerial vision proposed automatically analyze bolts’ condition. The contributions as follows: (1) Addressing problems motion blur often exists in videos captured by ariel systems (UASs) with high moving speed, bolt damage...

Unmanned aerial systems (UASs) are increasingly applied for bridge inspection. A vision-guided UAS with a lightweight convolutional neural network is developed to detect and locate cracks, spalling, corrosion. The contributions as follows: (1) To address the problem that traditional UASs global positioning system (GPS) required while GPS signals under bottom generally weak. designed applied, in which stereo vision-inertial fusion method used provide position data instead of an ultrasonic...

Accurate and robust ultrasound image segmentation is critical for computer-aided diagnostic systems. Nevertheless, the inherent challenges of imaging, such as blurry boundaries speckle noise, often cause traditional methods to struggle with performance. Despite recent advancements in universal segmentation, Segment Anything Model, existing interactive still suffer from inefficiency lack specialization. These rely heavily on extensive accurate manual or random sampling prompts interaction,...

Regular crack detection is essential for extending the service life of bridges. However, image data collected during bridge inspections are complex to convert into physical information and construct intuitive comprehensive Three-Dimensional (3D) models incorporating information. An intelligent method surface damage based on Unmanned Aerial Vehicles (UAVs) proposed these challenges, a three-stage detection, quantification, visualization process. This enables automatic localization in 3D...

Bridge deformation response data are the basis for calculating dynamic parameters of bridge, and it is great significance to accurately measure bridge during load test service conditions. A measurement method using an unmanned aerial system (UAS) with dual cameras a deep learning-based object tracking proposed deformation. The contributions as follows: (1) To address problem that movement UAS brings error results, telephoto wide-angle lenses used simultaneously capture deformed points stable...

Segmenting anatomical structures and lesions from ultrasound images contributes to disease assessment, diagnosis, treatment. Weakly supervised learning (WSL) based on sparse annotation has achieved encouraging performance demonstrated the potential reduce costs. However, often suffer issues such as poor contrast, unclear edges, well varying sizes locations of lesions. This makes it challenging for convolutional networks with local receptive fields extract global morphological features...

Abstract Damage inspection on the undersides of bridges is an important and challenging part routine bridge inspections. A method for 3D reconstruction damage localization based close-range photography by unmanned aerial vehicle (UAV) stereo vision combined with deep learning algorithms proposed, specific contributions include: (1) proposing a acquiring high-resolution images from multiple perspectives underside UAVs, serving as data source analysis; (2) applying learning-assisted...

Abstract Rapid inspection of urban road cracks is vital to maintain traffic smoothness and ensure safety. A rapid pavement crack method uses low-altitude aerial images captured by an unmanned system (UAS) deep-learning aided 3D reconstruction, a learning-based object segmentation algorithm proposed measure automatically. The contributions include: (a) An efficient reconstruction for UAS proposed, which applies instance network segment targets from raw with complex backgrounds first then...

Recently, the medical image segmentation have made rapid progress. Specifically, precision of play a pivotal role in realm disease diagnosis and treatment. Therefore, it is vital to improve performance. Generally, Transformer-based methods exhibit superior performance compared CNN-based on 3D tasks due their inherent capability capture global-aware context. However, existing transformer-based models are still unsatisfactory accuracy. In this paper, we propose novel fully convolution...

Abstract Accurate and efficient positioning is critical to ensuring the dimensional quality assessment of embedded steel plates. However, traditional manual measurement methods struggle efficiently measure evaluate these Vision-based offer advantages such as high resolution, fast data acquisition, processing speed, allowing accurate 2D coordinates. LiDAR can capture highly point clouds, due unordered nature analysis require significant computational resources. This paper proposes a method...

Segmenting internal structure from echocardiography is essential for the diagnosis and treatment of various heart diseases. Semi-supervised learning shows its ability in alleviating annotations scarcity. While existing semi-supervised methods have been successful image segmentation across medical imaging modalities, few attempted to design specifically addressing challenges posed by poor contrast, blurred edge details noise echocardiography. These characteristics pose generation high-quality...