Search Results (16,883)

General face recognition is currently one of the key technologies in the field of computer vision, and it has achieved tremendous success with the support of deep-learning technology. General face recognition models currently exhibit extremely high accuracy on some high-quality face datasets. However, their performance decreases in challenging environments, such as low-light scenes. To enhance the performance of face recognition models in low-light scenarios, we propose a face recognition approach based on feature decoupling and fusion (DeFFace). Our main idea is to extract facial-related features from images that are not influenced by illumination. First, we introduce a feature decoupling network (D-Net) to decouple the image into facial-related features and illumination-related features. By incorporating the illumination triplet loss optimized with unpaired identity IDs, we regulate illumination-related features to minimize the impact of lighting conditions on the face recognition system. However, the decoupled features are relatively coarse. Therefore, we introduce a feature fusion network (F-Net) to further extract the residual facial-related features from the illumination-related features and fuse them with the initial facial-related features. Finally, we introduce a lighting-facial correlation loss to reduce the correlation between the two decoupled features in the specific space. We demonstrate the effectiveness of our method on four real-world low-light datasets and three simulated low-light datasets. We retrain multiple general face recognition methods using our proposed low-light training sets to further validate the advanced performance of our method. Compared to general face recognition methods, our approach achieves an average improvement of more than 2.11 percentage points on low-light face datasets. In comparison with image enhancement-based solutions, our method shows an average improvement of around 16 percentage points on low-light datasets, and it also delivers an average improvement of approximately 5.67 percentage points when compared to illumination normalization-based methods. Full article

Debonding of the primary anode caused by anodic acidification is one of the major failure modes of the impressed current cathodic protection (ICCP) system in reinforced concrete structures. This study used 3D micro X-ray computed tomography (μ-XCT) to monitor the in situ evolution of the anodic acidification-affected zone. Samples were scanned after 0 to 40 days of the accelerated anodic acidification test. The anodic acidification-affected zone was identified in μ-XCT images using the gray level segmentation method. The total volume of this zone was measured using the 3D reconstruction method. It was found that detailed 3D information can be extracted using the 3D reconstruction method. The spatial heterogeneity was analyzed using this reconstructed volume information. The Faraday efficiency was calculated and found to increase after 20 days of operation. It was also found that the affected zone was proportional to the input electrical energy. The proposed model is useful for estimating the durability of an ICCP system. Full article

The rapid and accurate detection of fire scenes in various environments is crucial for effective disaster management and mitigation. Fire scene classification is a critical aspect of modern fire detection systems that directly affects public safety and property preservation. This research introduced a novel hybrid deep learning model designed to enhance the accuracy and efficiency of fire scene classification across diverse environments. The proposed model integrates advanced convolutional neural networks with multiscale feature extraction, attention mechanisms, and ensemble learning to achieve superior performance in real-time fire detection. By leveraging the strengths of pre-trained networks such as ResNet50, VGG16, and EfficientNet-B3, the model captures detailed features at multiple scales, ensuring robust detection capabilities. Including spatial and channel attention mechanisms further refines the focus on critical areas within the input images, reducing false positives and improving detection precision. Extensive experiments on a comprehensive dataset encompassing wildfires, building fires, vehicle fires, and non-fire scenes demonstrate that the proposed framework outperforms existing cutting-edge techniques. The model also exhibited reduced computational complexity and enhanced inference speed, making it suitable for deployment in real-time applications on various hardware platforms. This study sets a new benchmark for fire detection and offers a powerful tool for early warning systems and emergency response initiatives. Full article

The aim of this study was to explore the application of computational models for the analysis of histopathological images in the context of colon cancer. A comprehensive dataset of colon cancer images annotated into eight distinct categories based on their representation of cancerous cell portions was used. The primary objective was to employ various image classification algorithms to assess their efficacy in the context of cancer classification. Additionally, this study investigated the use of feature extraction techniques to derive meaningful data from the images, contributing to a more nuanced understanding of cancerous tissues, comparing the performance of different image classification algorithms in the context of colon cancer image analysis. The findings of this research suggested that XGboost provides the highest accuracy (89.79%) and could contribute to the growing body of knowledge in computational pathology. Other algorithms, such as the random forest, SVM, and CNN, also provided satisfactory results, offering insights into the effectiveness of image classification algorithms in distinguishing between different categories of cancerous cells. This work holds implications for the development of more accurate and efficient tools, underscoring the potential of computational models in enhancing the analysis of histopathological images and improving diagnostic capabilities in cancer research. Full article

Detecting cracks in cylinder heads traditionally relies on manual inspection, which is time-consuming and susceptible to human error. As an alternative, automated object detection utilizing computer vision and machine learning models has been explored. However, these methods often face challenges due to a lack of sufficiently annotated training data, limited image diversity, and the inherently small size of cracks. Addressing these constraints, this paper introduces a novel automated crack-detection method that enhances data availability through a synthetic data generation technique. Unlike general data augmentation practices, our method involves copying cracks from one location to another, guided by both random and informed engineering decisions about likely crack formations due to cyclic thermomechanical loads. The innovative aspect of our approach lies in the integration of domain-specific engineering knowledge into the synthetic generation process, which substantially improves detection accuracy. We evaluate our method’s effectiveness using two metrics: the F2 score, which emphasizes recall to prioritize detecting all potential cracks, and mean average precision (MAP), a standard measure in object detection. Experimental results demonstrate that, without engineering insights, our method increases the F2 score from 0.40 to 0.65, while maintaining a stable MAP. Incorporating detailed engineering knowledge further enhances the F2 score to 0.70 and improves MAP to 0.57, representing increases of 63% and 43%, respectively. These results confirm that our approach not only mitigates the limitations of traditional data augmentation but also significantly advances the reliability and precision of crack detection in industrial settings. Full article

To address the challenge of rapid geometric model development in the digital twin industry, this paper presents a comprehensive pipeline for constructing 3D models from images using monocular vision imaging principles. Firstly, a structure-from-motion (SFM) algorithm generates a 3D point cloud from photographs. The feature detection methods scale-invariant feature transform (SIFT), speeded-up robust features (SURF), and KAZE are compared across six datasets, with SIFT proving the most effective (matching rate higher than 0.12). Using K-nearest-neighbor matching and random sample consensus (RANSAC), refined feature point matching and 3D spatial representation are achieved via antipodal geometry. Then, the Poisson surface reconstruction algorithm converts the point cloud into a mesh model. Additionally, texture images are enhanced by leveraging a visual geometry group (VGG) network-based deep learning approach. Content images from a dataset provide geometric contours via higher-level VGG layers, while textures from style images are extracted using the lower-level layers. These are fused to create texture-transferred images, where the image quality assessment (IQA) metrics SSIM and PSNR are used to evaluate texture-enhanced images. Finally, texture mapping integrates the enhanced textures with the mesh model, improving the scene representation with enhanced texture. The method presented in this paper surpassed a LiDAR-based reconstruction approach by $20\%$ in terms of point cloud density and number of model facets, while the hardware cost was only $1\%$ of that associated with LiDAR. Full article

Lung and colon cancers are among the leading causes of cancer-related mortality worldwide. Early and accurate detection of these cancers is crucial for effective treatment and improved patient outcomes. False or incorrect detection is harmful. Accurately detecting cancer in a patient’s tissue is crucial to their effective treatment. While analyzing tissue samples is complicated and time-consuming, deep learning techniques have made it possible to complete this process more efficiently and accurately. As a result, researchers can study more patients in a shorter amount of time and at a lower cost. Much research has been conducted to investigate deep learning models that require great computational ability and resources. However, none of these have had a $100\%$ accurate detection rate for these life-threatening malignancies. Misclassified or falsely detecting cancer can have very harmful consequences. This research proposes a new lightweight, parameter-efficient, and mobile-embedded deep learning model based on a 1D convolutional neural network with squeeze-and-excitation layers for efficient lung and colon cancer detection. This proposed model diagnoses and classifies lung squamous cell carcinomas and adenocarcinoma of the lung and colon from digital pathology images. Extensive experiment demonstrates that our proposed model achieves $100\%$ accuracy for detecting lung, colon, and lung and colon cancers from the histopathological (LC25000) lung and colon datasets, which is considered the best accuracy for around $0.35$ million trainable parameters and around $6.4$ million flops. Compared with the existing results, our proposed architecture shows state-of-the-art performance in lung, colon, and lung and colon cancer detection. Full article

Sphere formation assay is an accepted cancer stem cell (CSC) enrichment method. CSCs play a crucial role in chemoresistance and cancer recurrence. Therefore, CSC growth is studied in plates and microdevices to develop prediction chemotherapy assays in cancer. As counting spheres cultured in devices is laborious, time-consuming, and operator-dependent, a computational program called the Automatic Quantification of Spheres Algorithm (ASQA) that detects, identifies, counts, and measures spheres automatically was developed. The algorithm and manual counts were compared, and there was no statistically significant difference (p = 0.167). The performance of the AQSA is better when the input image has a uniform background, whereas, with a nonuniform background, artifacts can be interpreted as spheres according to image characteristics. The areas of spheres derived from LN229 cells and CSCs from primary cultures were measured. For images with one sphere, area measurements obtained with the AQSA and SpheroidJ were compared, and there was no statistically significant difference between them (p = 0.173). Notably, the AQSA detects more than one sphere, compared to other approaches available in the literature, and computes the sphere area automatically, which enables the observation of treatment response in the sphere derived from the human glioblastoma LN229 cell line. In addition, the algorithm identifies spheres with numbers to identify each one over time. The AQSA analyzes many images in 0.3 s per image with a low computational cost, enabling laboratories from developing countries to perform sphere counts and area measurements without needing a powerful computer. Consequently, it can be a useful tool for automated CSC quantification from cancer cell lines, and it can be adjusted to quantify CSCs from primary culture cells. CSC-derived sphere detection is highly relevant as it avoids expensive treatments and unnecessary toxicity. Full article

Automatic clothes pattern recognition is important to assist visually impaired people and for real-world applications such as e-commerce or personal fashion recommendation systems, and it has attracted increased interest from researchers. It is a challenging texture classification problem in that even images of the same texture class expose a high degree of intraclass variations. Moreover, images of clothes patterns may be taken in an unconstrained illumination environment. Machine learning methods proposed for this problem mostly rely on handcrafted features and traditional classification methods. The research works that utilize the deep learning approach result in poor recognition performance. We propose a deep learning method based on an ensemble of convolutional neural networks where feature engineering is not required while extracting robust local and global features of clothes patterns. The ensemble classifier employs a pre-trained ResNet50 with a non-local (NL) block, a squeeze-and-excitation (SE) block, and a coordinate attention (CA) block as base learners. To fuse the individual decisions of the base learners, we introduce a simple and effective fusing technique based on entropy voting, which incorporates the uncertainties in the decisions of base learners. We validate the proposed method on benchmark datasets for clothes patterns that have six categories: solid, striped, checkered, dotted, zigzag, and floral. The proposed method achieves promising results for limited computational and data resources. In terms of accuracy, it achieves 98.18% for the GoogleClothingDataset and 96.03% for the CCYN dataset. Full article

This work presents a novel RGB-D dynamic simultaneous localization and mapping (SLAM) method that improves accuracy, stability, and efficiency of localization while relying on deep learning in a dynamic environment, in contrast to traditional static scene-based visual SLAM methods. Based on the classic framework of traditional visual SLAM, we propose a method that replaces the traditional feature extraction method with a convolutional neural network approach, aiming to enhance the accuracy of feature extraction and localization, as well as to improve the algorithm’s ability to capture and represent the characteristics of the entire scene. Subsequently, the semantic segmentation thread was utilized in a target detection network combined with geometric methods to identify potential dynamic areas in the image and generate masks for dynamic objects. Finally, the standard deviation of the depth information of potential dynamic points was calculated to identify true dynamic feature points, to guarantee that static feature points were used for position estimation. We performed experiments based on the public datasets to validate the feasibility of the proposed algorithm. The experimental results indicate that the improved SLAM algorithm, which boasts a reduction in absolute trajectory error (ATE) by approximately 97% compared to traditional static visual SLAM and about 20% compared to traditional dynamic visual SLAM, also exhibited a 68% decrease in computation time compared to well-known dynamic visual SLAM, thereby possessing absolute advantages in both positioning accuracy and operational efficiency. Full article

The multiphase components of natural ice contain gas, ice, unfrozen water, sediment and brine. X-ray computed tomography (CT) analysis of ice multiphase components has the advantage of high precision, non-destructiveness and visualization; however, it is limited by the segmentation thresholds. Due to the proximity of the CT value ranges of gas, ice, unfrozen water, sediment and brine within the samples, there is uncertainty in the artificial determination of the CT image segmentation thresholds, as well as unsuitability of the global threshold segmentation methods. In order to improve the accuracy of multi-threshold segmentation in CT images, a CT system was used to scan the Yellow River ice, the Wuliangsuhai lake ice and the Arctic sea ice. The threshold ranges of multiphase components within the ice were determined by watershed algorithm to construct a high-precision three-dimensional ice model. The results indicated that CT combined with watershed algorithm was an efficient and non-destructive method for obtaining microscopic information within ice, which accurately segmented the ice into multiphase components such as gas, ice, unfrozen water, sediment, and brine. The gas CT values of the Yellow River ice, the Wuliangsuhai lake ice and the Arctic sea ice ranged from −1024 Hu~−107 Hu, −1024 Hu~−103 Hu, and −1024 Hu~−160 Hu, respectively. The ice CT values of the Yellow River ice, the Wuliangsuhai lake ice and the Arctic sea ice ranged from −103 Hu~−50 Hu, −100 Hu~−38 Hu, −153 Hu~−51 Hu. The unfrozen water CT values of the Yellow River ice and the Wuliangsuhai lake ice ranged from −8 Hu~18 Hu, −8 Hu~13 Hu. The sediment CT values of the Yellow River ice and the Wuliangsuhai lake ice ranged from 20 Hu~3071 Hu, 20 Hu~3071 Hu, and the brine CT values of the Arctic sea ice ranged from −6 Hu~3071 Hu. The errors between the three-dimensional ice model divided by threshold ranges and measured sediment content were less than 0.003 g/cm³, which verified the high accuracy of the established microscopic model. It provided a scientific basis for ice engineering, ice remote sensing, and ice disaster prevention. Full article

Accurate 3D object detection is essential for autonomous driving, yet traditional LiDAR models often struggle with sparse point clouds. We propose perspective-aware hierarchical vision transformer-based LiDAR-camera fusion (PLC-Fusion) for 3D object detection to address this. This efficient, multi-modal 3D object detection framework integrates LiDAR and camera data for improved performance. First, our method enhances LiDAR data by projecting them onto a 2D plane, enabling the extraction of object perspective features from a probability map via the Object Perspective Sampling (OPS) module. It incorporates a lightweight perspective detector, consisting of interconnected 2D and monocular 3D sub-networks, to extract image features and generate object perspective proposals by predicting and refining top-scored 3D candidates. Second, it leverages two independent transformers—CamViT for 2D image features and LidViT for 3D point cloud features. These ViT-based representations are fused via the Cross-Fusion module for hierarchical and deep representation learning, improving performance and computational efficiency. These mechanisms enhance the utilization of semantic features in a region of interest (ROI) to obtain more representative point features, leading to a more effective fusion of information from both LiDAR and camera sources. PLC-Fusion outperforms existing methods, achieving a mean average precision (mAP) of 83.52% and 90.37% for 3D and BEV detection, respectively. Moreover, PLC-Fusion maintains a competitive inference time of 0.18 s. Our model addresses computational bottlenecks by eliminating the need for dense BEV searches and global attention mechanisms while improving detection range and precision. Full article

Lower-limb alignment deformities constitute a significant clinical concern, as they can lead to serious complications, including progressive degenerative diseases and disabilities. Rotational deformities may give rise to conditions such as joint arthrosis, patellar instability, and the degeneration of the patellofemoral cartilage. Therefore, a comprehensive evaluation of lower-limb alignment is essential for the effective patient management, preoperative planning, and successful correction of these deformities. The primary assessment method employs full-length standing radiographs in the anteroposterior (AP) projection, which facilitates accurate measurements of the anatomical and mechanical axes of the lower limb, including angles and deviations. The outcomes of this analysis are vital for the meticulous planning of osteotomy and total knee arthroplasty (TKA). In addition, computed tomography (CT) provides a specialized approach for the precise evaluation of femoral and tibial rotation. In this area, there are potential opportunities for the implementation of AI-based automated measurement systems. Full article

Atmospheric visibility is a crucial meteorological element impacting urban air pollution monitoring, public transportation, and military security. Traditional visibility detection methods, primarily manual and instrumental, have been costly and imprecise. With advancements in data science and computing, deep learning-based visibility detection technologies have rapidly emerged as a research hotspot in atmospheric science. This paper systematically reviews the applications of various deep learning models—Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Generative Adversarial Networks (GANs), and Transformer networks—in visibility estimation, prediction, and enhancement. Each model’s characteristics and application methods are discussed, highlighting the efficiency of CNNs in spatial feature extraction, RNNs in temporal tracking, GANs in image restoration, and Transformers in capturing long-range dependencies. Furthermore, the paper addresses critical challenges in the field, including dataset quality, algorithm optimization, and practical application barriers, proposing future research directions, such as the development of large-scale, accurately labeled datasets, innovative learning strategies, and enhanced model interpretability. These findings highlight the potential of deep learning in enhancing atmospheric visibility detection techniques, providing valuable insights into the literature and contributing to advances in the field of meteorological observation and public safety. Full article

Neural Architecture Search (NAS) has found applications in various areas of computer vision, including image recognition and object detection. An increasing number of algorithms, such as ENAS (Efficient Neural Architecture Search via Parameter Sharing) and DARTS (Differentiable Architecture Search), have been applied to NAS. Nevertheless, the current Training-free NAS methods continue to exhibit unreliability and inefficiency. This paper introduces a training-free prune-based algorithm called TTNAS (True-Skill Training-Free Neural Architecture Search), which utilizes a Bayesian method (true-skill algorithm) to combine multiple indicators for evaluating neural networks across different datasets. The algorithm demonstrates highly competitive accuracy and efficiency compared to state-of-the-art approaches on various datasets. Specifically, it achieves 93.90% accuracy on CIFAR-10, 71.91% accuracy on CIFAR-100, and 44.96% accuracy on ImageNet 16-120, using 1466 GPU seconds in NAS-Bench-201. Additionally, the algorithm exhibits improved adaptation to other datasets and tasks. Full article