Vehicle Pedestrian Detection Method Based on Spatial Pyramid Pooling and Attention Mechanism

In order to solve the stereo matching problem better, the accuracy of the algorithm in the stereo matching task needs to be improved. To tackle the problem, we proposed a Dual Spatial Pyramid Pooling Stereo Matching Network with Attention Mechanism (DSANet), a stereo matching network with better multi-scale feature extraction capability based on the basic structure of Pyramid Stereo Matching Network (PSMNet). The network consists of two main modules: the dual spatial pyramid pooling module and the attention module. The former module extracts multi-scale features from different branches at the same time. After concatenating these different features, the latter module will optimize the fused features and generate the cost volume with different branches multi-scale feature. The proposed approach was evaluated on the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) 2015 dataset. Our experiment results on the KITTI 2015 dataset show that the mismatch rate of our network is 3.02%. Compared with the basic structure of PSMNet, the proposed method improves the effect of stereo matching.

Dynamic street-scene reconstruction from sparse viewpoints over long temporal spans is challenged by temporal instability, ghosting near occlusions, and background drift. This paper presents SPT-Gauss (Semantic Prior and Temporal constraint-enhanced Gaussian splatting), a Gaussian-splatting framework that improves dynamic reconstruction without object-level annotations by combining dense semantic priors with lightweight, parameter-level temporal regularization. SPT-Gauss distills per-pixel semantic features from a frozen 2D foundation model into 4D Gaussian primitives, estimates static and dynamic regions via a dual-evidence motion mask, and regularizes temporal parameters through a semantic-guided velocity constraint and a static-lifetime prior to suppress spurious background motion. Experiments on the Waymo Open Dataset and KITTI (Karlsruhe Institute of Technology and Toyota Technological Institute) show consistent improvements over representative baselines in both 4D reconstruction and novel-view synthesis, with reduced temporal artifacts and improved fidelity in motion-challenging regions.

Background: 3D object detection based on point clouds in road scenes has attracted much attention recently. The voxel-based methods voxelize the scene to regular grids, which can be processed with the advanced feature learning frameworks based on convolutional layers for semantic feature learning. The point-based methods can extract the geometric feature of the point due to the coordinate reservations. The combination of the two is effective for 3D object detection. However, the current methods use a voxel-based detection head with anchors for classification and localization. Although the preset anchors cover the entire scene, it is not suitable for detection tasks with larger scenes and multiple categories of objects, due to the limitation of the voxel size. Additionally, the misalignment between the predicted confidence and proposals in the Regions of the Interest (ROI) selection bring obstacles to 3D object detection. Methods: We investigate the combination of voxel-based methods and point-based methods for 3D object detection. Additionally, a voxel-to-point module that captures semantic and geometric features is proposed in the paper. The voxel-to-point module is conducive to the detection of small-size objects and avoids the presets of anchors in the inference stage. Moreover, a confidence adjustment module with the center-boundary-aware confidence attention is proposed to solve the misalignment between the predicted confidence and proposals in the regions of the interest selection. Results: The proposed method has achieved state-of-the-art results for 3D object detection in the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) object detection dataset. Actually, as of September 19, 2021, our method ranked 1st in the 3D and Bird Eyes View (BEV) detection of cyclists tagged with difficulty level ‘easy’, and ranked 2nd in the 3D detection of cyclists tagged with ‘moderate’. Conclusions: We propose an end-to-end two-stage 3D object detector with voxel-to-point module and confidence adjustment module.

YOLOv3 is a deep learning-based real-time object detector and is mainly used in applications such as video surveillance and autonomous vehicles. In this paper, we proposed an improved YOLOv3 (You Only Look Once version 3) applied Duplex FPN, which enhanced large object detection by utilizing low-level feature information. The conventional YOLOv3 improved the small object detection performance by applying FPN (Feature Pyramid Networks) structure to YOLOv2. However, YOLOv3 with an FPN structure specialized in detecting small objects, so it is difficult to detect large objects. Therefore, this paper proposed an improved YOLOv3 applied Duplex FPN, which can utilize low-level location information in high-level feature maps instead of the existing FPN structure of YOLOv3. This improved the detection accuracy of large objects. Also, an extra detection layer was added to the top-level feature map to prevent failure of detection of parts of large objects. Further, dimension clusters of each detection layer were reassigned to learn quickly how to accurately detect objects. The proposed method was compared and analyzed in the PASCAL VOC dataset. The experimental results showed that the bounding box accuracy of large objects improved owing to the Duplex FPN and extra detection layer, and the proposed method succeeded in detecting large objects that the existing YOLOv3 did not.

Object detection and recognition is an essential task in computer vision with numerous real-world applications such as surveillance, self-driving cars, and robotics. In recent years, deep learning-based approaches have significantly improved the accuracy and speed of object detection and recognition. The You Only Look Once version 3 (YOLOv3) algorithm is a popular deep learning-based approach that can detect and recognize objects in real-time. The Common Objects in Context (COCO) dataset is a large-scale dataset with over 330,000 labeled images and more than 2.5 million object instances, making it a popular choice for object detection and recognition tasks. In this paper, we propose a deep learning-based approach for real-time object detection and recognition using the YOLOv3 architecture and COCO dataset. We evaluate our approach based on several performance metrics, including mean average precision (mAP), frames per second (FPS), total object detection time, object detection accuracy, false positive rate, number of detected objects, and mean intersection over union (mIoU). Our results show that our approach achieves a mean average precision of 0.76 on the COCO dataset and a real-time performance of 40 frames per second on a single GPU. Additionally, our approach achieves an object detection accuracy of 93.5%, a false positive rate of 6.5%, and a mean intersection over union of 0.65. Our proposed approach shows promising results for real-time object detection and recognition and can be applied to various real-world applications.

Abstract. Target detection is a vital field within computer vision, playing an essential role in applications. This paper investigates the advancements and efficiencies of contemporary target detection methodologies, focusing on deep learning frameworks. Through a systematic review and evaluation of these models' architectures and performances using the Common Objects in Context (COCO) dataset, the study highlights their operational effectiveness and practical implications in real-world scenarios. A detailed comparative analysis is conducted, assessing the models based on mean Average Precision (mAP) and input dimensions to determine their suitability across various detection tasks. You Only Look Once version 3 (YOLOv3), in particular, is recognized for its ability to combine high-speed detection with significant accuracy, effectively addressing real-time processing challenges. The results confirm that YOLOv3, despite its smaller input size, performs comparably to more complex systems, demonstrating notable enhancements in design efficiency and processing speed. This research underscores the potential for future optimizations in model architectures to bridge the gap between high-speed and high-accuracy detection tasks, potentially transforming real-time applications across multiple sectors. The practical significance of this work lies in its ability to guide future developments in object detection, benefiting both academic research and industrial applications.

Recently, deep learning methods and multisensory fusion have been applied to address odometry challenges in unmanned ground vehicles (UGVs). In this paper, we propose an end-to-end visual-lidar-inertial odometry framework to enhance the accuracy of pose estimation. Grayscale images, 3D point clouds, and inertial data are used as inputs to overcome the limitations of a single sensor. Convolutional neural network (CNN) and recurrent neural network (RNN) are employed as encoders for different sensor modalities. In contrast to previous multisensory odometry methods, our framework introduces a novel attention-based fusion module that remaps feature vectors to adapt to various scenes. Evaluations on the Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago (KITTI) odometry benchmark demonstrate the effectiveness of our framework.

Unmanned Aerial Vehicles (UAVs) have gained significant traction for real-time surveillance applications such as object tracking and search-and-rescue missions. A key enabler of these capabilities is the integration of a real-time, highly accurate object detection system. While available two-stage detectors like R-CNN, Fast R-CNN, Faster R-CNN, and Mask R-CNN achieve high detection accuracy and precise localization by dividing the image into regions and classifying each region, they are often slow, complex, and resource-intensive, making them unsuitable for real-time applications. To address this limitation, this research utilizes a Darknet-based YOLOv3 (You Only Look Once, Version 3), a state-of-the-art algorithm for real-time object detection in videos, live feeds, and images. Leveraging a deep Convolutional Neural Network (CNN), YOLOv3 efficiently learns features and predicts object locations and class probabilities in a single pass, ensuring high-speed and reliable detection. A real-time aerial surveillance system for quadcopter application using machine vision is proposed. The training dataset, obtained from the internet and self-taken images from a camera, was manually annotated into three critical categories: suspect, bandit, and weapon. The dataset was subsequently divided into training and testing subsets. Experimental results demonstrate that the proposed system achieves outstanding detection accuracy, with an overall mean average precision (mAP@0.5) of 93.82%, precision of 94%, and recall of 83%. Compared to a ResNet-50-based Faster R-CNN model, the YOLOv3-based approach outperformed, achieving success rates of 0.81 for the Bandit class and 1.00 for both Suspect and Weapon classes. This research improves drone-based AI for real-time object detection in security systems, enhancing efficiency and adaptability. It also creates specialized datasets, including a tailored bandit dataset, to support future UAV security research.

Radiographic interpretation suffers from an ever-increasing workload in orthopedic and radiology departments. The present study applied and assessed the performance of a convolutional neural network designed to assist orthopedists and radiologists in the detection and classification of knee osteoarthritis from early to severe degrees in accordance with the Kellgren-Lawrence (KL) classification system. In total, 1650 knee joint radiographs (anteroposterior view) were collected from the Osteoarthritis Initiative public resource. Two models were developed: one distinguished normal (KL 0-I) from osteoarthritic knees (KL II-IV), and the other classified the severity as normal (KL 0-I), non-severe (KL II), or severe (KL III-IV). The regions of interest were labeled under the supervision of experts. Our artificial intelligence (AI) models were trained using the You Only Look Once version 3 (YOLOv3) detection algorithm. Our first AI model using YOLOv3 tiny could detect and classify normal and osteoarthritic knees on plain knee joint radiographs with 85% accuracy and 81% mean average precision. The second AI model for classifying severity achieved a total accuracy of 86.7% and mean average precision of 70.6%. Our proposed deep learning models provided high accuracy and satisfactory precision for the detection and classification of early to severe knee osteoarthritis on anteroposterior radiographs. These models may be used as diagnostic aids by interpreting knee radiographs and guiding the treatment options via each osteoarthritic stage for related physicians and specialists.

In unstructured environments, robots face challenges in efficiently and accurately grasping irregular, fragile objects. To address this, this paper introduces a soft robotic hand tailored for such settings and enhances You Only Look Once v5s (YOLOv5s), a lightweight detection algorithm, to achieve efficient grasping. A rapid pneumatic network-based soft finger structure, broadly applicable to various irregularly placed objects, is designed, with a mathematical model linking the bending angle of the fingers to input gas pressure, validated through simulations. The YOLOv5s model is improved by integrating the Coordinate Attention (CA) mechanism in the backbone layer, refining the Spatial Pyramid Pooling (SPP) module for faster detection, and adjusting the loss function to prevent misalignment between predicted and actual bounding boxes, thereby enhancing computational efficiency. Experimental comparative analysis indicates that the refined model exhibits improvements in both mean average precision and recognition speed. A soft robotic grasping experimental platform was established, with precision grasping and power grasping experiments conducted using the pose and object type data generated by the enhanced YOLOv5s-CA-SPP model network. The results show that the success rate of grabbing reaches 82% with a proper grabbing posture.

With the rapid expansion of railway networks globally, ensuring rail infrastructure safety through efficient detection methods has become critical. Traditional inspection systems face limitations in flexibility, adaptability to adverse weather, and multifunctional integration. This study proposes a ground-air collaborative multi-source detection system that integrates 3D light detection and ranging (LiDAR)-based point cloud imaging and deep learning-driven intrusion detection. The system employs a lightweight rail inspection vehicle equipped with dual LiDARs and an Astro camera, synchronized with an unmanned aerial vehicle (UAV) carrying industrial-grade LiDAR. We propose an improved LiDAR odometry and mapping with sliding window (LOAM-SLAM) algorithm enables real-time dynamic mapping, while an optimized iterative closest point (ICP) algorithm achieves high-precision point cloud registration and colorization. For intrusion detection, a You Only Look Once version 3 (YOLOv3)-ResNet fusion model achieves a recall rate of 0.97 and precision of 0.99. The system's innovative design and technical implementation offer significant improvements in railway track inspection efficiency and safety. This work establishes a new paradigm for adaptive railway maintenance in complex environments.

We propose an object detection method that improves the accuracy of the conventional SSD (Single Shot Multibox Detector), which is one of the top object detection algorithms in both aspects of accuracy and speed. The performance of a deep network is known to be improved as the number of feature maps increases. However, it is difficult to improve the performance by simply raising the number of feature maps. In this paper, we propose and analyze how to use feature maps effectively to improve the performance of the conventional SSD. The enhanced performance was obtained by changing the structure close to the classifier network, rather than growing layers close to the input data, e.g., by replacing VGGNet with ResNet. The proposed network is suitable for sharing the weights in the classifier networks, by which property, the training can be faster with better generalization power. For the Pascal VOC 2007 test set trained with VOC 2007 and VOC 2012 training sets, the proposed network with the input size of 300 × 300 achieved 78.5% mAP (mean average precision) at the speed of 35.0 FPS (frame per second), while the network with a 512 × 512 sized input achieved 80.8% mAP at 16.6 FPS using Nvidia Titan X GPU. The proposed network shows state-of-the-art mAP, which is better than those of the conventional SSD, YOLO, Faster-RCNN and RFCN. Also, it is faster than Faster-RCNN and RFCN.

As a sub-direction of image retrieval, person re-identification (Re-ID) is usually used to solve the security problem of cross camera tracking and monitoring. A growing number of shopping centers have recently attempted to apply Re-ID technology. One of the development trends of related algorithms is using an attention mechanism to capture global and local features. We notice that these algorithms have apparent limitations. They only focus on the most salient features without considering certain detailed features. People’s clothes, bags and even shoes are of great help to distinguish pedestrians. We notice that global features usually cover these important local features. Therefore, we propose a dual branch network based on a multi-scale attention mechanism. This network can capture apparent global features and inconspicuous local features of pedestrian images. Specifically, we design a dual branch attention network (DBA-Net) for better performance. These two branches can optimize the extracted features of different depths at the same time. We also design an effective block (called channel, position and spatial-wise attention (CPSA)), which can capture key fine-grained information, such as bags and shoes. Furthermore, based on ID loss, we use complementary triplet loss and adaptive weighted rank list loss (WRLL) on each branch during the training process. DBA-Net can not only learn semantic context information of the channel, position, and spatial dimensions but can integrate detailed semantic information by learning the dependency relationships between features. Extensive experiments on three widely used open-source datasets proved that DBA-Net clearly yielded overall state-of-the-art performance. Particularly on the CUHK03 dataset, the mean average precision (mAP) of DBA-Net achieved 83.2%.

.Significance: Automatic and accurate classification of three-dimensional (3-D) retinal optical coherence tomography (OCT) images is essential for assisting ophthalmologist in the diagnosis and grading of macular diseases. Therefore, more effective OCT volume classification for automatic recognition of macular diseases is needed.Aim: For OCT volumes in which only OCT volume-level labels are known, OCT volume classifiers based on its global feature and deep learning are designed, validated, and compared with other methods.Approach: We present a general framework to classify OCT volume for automatic recognizing macular diseases. The architecture of the framework consists of three modules: B-scan feature extractor, two-dimensional (2-D) feature map generation, and volume-level classifier. Our architecture could address OCT volume classification using two 2-D image machine learning classification algorithms. Specifically, a convolutional neural network (CNN) model is trained and used as a B-scan feature extractor to construct a 2-D feature map of an OCT volume and volume-level classifiers such as support vector machine and CNN with/without attention mechanism for 2-D feature maps are described.Results: Our proposed methods are validated on the publicly available Duke dataset, which consists of 269 intermediate age-related macular degeneration (AMD) volumes and 115 normal volumes. Fivefold cross-validation was done, and average accuracy, sensitivity, and specificity of 98.17%, 99.26%, and 95.65%, respectively, are achieved. The experiments show that our methods outperform the state-of-the-art methods. Our methods are also validated on our private clinical OCT volume dataset, consisting of 448 AMD volumes and 462 diabetic macular edema volumes.Conclusions: We present a general framework of OCT volume classification based on its 2-D feature map and CNN with attention mechanism and describe its implementation schemes. Our proposed methods could classify OCT volumes automatically and effectively with high accuracy, and they are a potential practical tool for screening of ophthalmic diseases from OCT volume.

Reliable evaluations of the radiological consequences of hypothetical severe accidents in nuclear power plants (NPPs) play a key role in supporting the emergency and preparedness plans during such abnormal events. Also triggered by the growing requests of the national regulators, one of the milestones of the Nuclear Safety Research program (NUSAFE) of the Karlsruhe Institute of Technology (KIT) is the improvement of the performance of the codes employed for such evaluations in the current and innovative NPPs, which are in operation and going to be built in EU and worldwide. Having this in mind, a calculation platform of state-of-art codes has been assessed at KIT to evaluate the accident progression, the radiological source term, and the behavior of the fission product dispersion to the environment by also taking into account the corresponding uncertainties. In particular, the CASMO5 lattice physics code, developed by Studsvik Scandpower, is used for a proper evaluation of the fuel inventory loaded in the reactor core. The European reference Accident Source Term Evaluation Code (ASTEC), developed by IRSN (France), is employed to evaluate the accident progression from the initiation to the fission produced release to the environment. The use of the KArlsruhe Tool for Uncertainty and Sensitivity Analysis (KATUSA), developed by KIT, aims at performing the uncertainty quantification of the source term. Finally, the Java Real-time On-line DecisiOn Support (JRODOS) is employed for analyzing the radiological consequences of the accident. The KIT platform has been employing since some years in the frame of the KIT participation to different European projects, i.e., MUSA, SASPAM-SA on SMRs, SANE. In the paper, an overall view of the KIT strategy on severe accident analyses will be provided, and examples of the application of the above-mentioned platform to generic NPPs will be provided, i.e., PWR Konvoi.