Dense Objects Research Articles

Automated Physical Distance Estimation and Crowd Monitoring Through Surveillance Video.

The contagious Corona Virus (COVID-19) transmission can be reduced by following and maintaining physical distancing (also known as COVID-19 social distance). The World Health Organisation (WHO) recommends preventing COVID-19 from spreading in public areas. On the other hand, people may not be maintaining the required 2-m physical distance as a mandated safety precaution in shopping malls and public places. The spread of the fatal disease may be slowed by an active monitoring system suitable for identifying distances between people and alerting them. This paper introduced a deep learning-based system for automatically detecting physical distance using video from security cameras. The proposed system introduced the TH-YOLOv5 for object detection and classification and Deepsort for tracking the detected people using bounding boxes from the video. TH-YOLOv5 included another prediction head to identify objects of varying sizes. The original prediction heads are then replaced with Transformer Heads (TH) to investigate the prediction capability of the self-attention mechanism. Then, we include the convolutional block attention model (CBAM) to identify attention areas in settings with dense objects. Pairwise L2 vectorized normalization was utilized to generate a three-dimensional feature space for tracking physical distances and the violation index, determining the number of individuals who follow the distance rules. We use the MS COCO and HumanCrowd, CityPersons, and Oxford Town Centre (OTC) data sets for training and testing. Experimental results demonstrate that the proposed system obtained a weighted mAP score of 89.5% and an FPS score of 29; both are computationally comparable.

Unified Analytic Melt-Shear Model in the Limit of Quantum Melting

Quantum melting is the phenomenon of cold (zero-temperature) melting of a pressure-ionized substance which represents a lattice of bare ions immersed in the background of free electrons, i.e., the so-called one-component plasma (OCP). It occurs when the compression of the substance corresponds to the zero-point fluctuations of its ions being so large that the ionic ordered state can no longer exist. Quantum melting corresponds to the classical melting curve reaching a turnaround point beyond which it starts going down and eventually terminates, when zero temperature is reached, at some critical density. This phenomenon, as well as the opposite phenomenon of quantum crystallization, may occur in dense stellar objects such as white dwarfs, and may play an important role in their evolution that requires a reliable thermoelasticity model for proper physical description. Here we suggest a modification of our unified analytic melt-shear thermoelasticity model in the region of quantum melting, and derive the corresponding Grüneisen parameters. We demonstrate how the new functional form for the cold shear modulus can be combined with a known equation of state. One of the constituents of the new model is the melting curve of OCP crystal which we also present. The inclusion of quantum melting implies that the modified model becomes applicable in the entire density range of the existence of the solid state, up to the critical density of quantum melting above which the solid state does not exist. Our approach can be generalized to model melting curves and cold shear moduli of different solid phases of a multi-phase material over the corresponding ranges of mechanical stability.

RD-YOLO: An Effective and Efficient Object Detector for Roadside Perception System.

In recent years, intelligent driving technology based on vehicle-road cooperation has gradually become a research hotspot in the field of intelligent transportation. There are many studies regarding vehicle perception, but fewer studies regarding roadside perception. As sensors are installed at different heights, the roadside object scale varies violently, which burdens the optimization of networks. Moreover, there is a large amount of overlapping and occlusion in complex road environments, which leads to a great challenge of object distinction. To solve the two problems raised above, we propose RD-YOLO. Based on YOLOv5s, we reconstructed the feature fusion layer to increase effective feature extraction and improve the detection capability of small targets. Then, we replaced the original pyramid network with a generalized feature pyramid network (GFPN) to improve the adaptability of the network to different scale features. We also integrated a coordinate attention (CA) mechanism to find attention regions in scenarios with dense objects. Finally, we replaced the original Loss with Focal-EIOU Loss to improve the speed of the bounding box regression and the positioning accuracy of the anchor box. Compared to the YOLOv5s, the RD-YOLO improves the mean average precision (mAP) by 5.5% on the Rope3D dataset and 2.9% on the UA-DETRAC dataset. Meanwhile, by modifying the feature fusion layer, the weight of RD-YOLO is decreased by 55.9% while the detection speed is almost unchanged. Nevertheless, the proposed algorithm is capable of real-time detection at faster than 71.9 frames/s (FPS) and achieves higher accuracy than the previous approaches with a similar FPS.

The sweep method for radiative transfer in arepo

ABSTRACT We introduce the radiative transfer code Sweep for the cosmological simulation suite arepo. Sweep is a discrete ordinates method in which the radiative transfer equation is solved under the infinite speed of light, steady state assumption by a transport sweep across the entire computational grid. Since arepo is based on an adaptive, unstructured grid, the dependency graph induced by the sweep dependencies of the grid cells is non-trivial. In order to solve the topological sorting problem in a distributed manner, we employ a task-based-parallelism approach. The main advantage of the sweep method is that the computational cost scales only with the size of the grid and is independent of the number of sources or the distribution of sources in the computational domain, which is an advantage for radiative transfer in cosmological simulations, where there are large numbers of sparsely distributed sources. We successfully apply the code to a number of physical tests such as the expansion of H ii regions, the formation of shadows behind dense objects, the scattering of light, and its behaviour in the presence of periodic boundary conditions. In addition, we measure its computational performance with a focus on highly parallel, large-scale simulations.

Fused x-ray and fast neutron CT reconstruction for imaging large and dense objects

Megavolt x-ray computed tomography (CT) is a powerful tool for three-dimensional characterization. However, its utility is limited for large objects composed of high-atomic number (Z) materials, where x rays fail to penetrate. Information from fast neutron CT (FNCT) can complement x-ray CT reconstructions since fast neutrons can more readily penetrate high-Z objects. Here, we demonstrate a method for combining FNCT and x-ray CT data to create a single reconstruction, more accurate than could be achieved with either x rays or fast neutrons alone. The algorithm was tested on an exemplar comprising multiple concentric, nested cylinders of different materials. Simulated and empirical x-ray CT data were acquired for the exemplar using a 9 MV bremsstrahlung spectrum. Additional simulated and empirical FNCT data were acquired using an accelerator based fast neutron source. The FNCT data were used to synthesize x-ray CT data and augment the x-ray CT data missing due to lack of penetration. This approach mitigates artifacts that would otherwise negatively affect the accuracy and resolution of a single-modality reconstructed volume.

Crowd Density Estimation in Spatial and Temporal Distortion Environment Using Parallel Multi-Size Receptive Fields and Stack Ensemble Meta-Learning

The estimation of crowd density is crucial for applications such as autonomous driving, visual surveillance, crowd control, public space planning, and warning visually distracted drivers prior to an accident. Having strong translational, reflective, and scale symmetry, models for estimating the density of a crowd yield an encouraging result. However, dynamic scenes with perspective distortions and rapidly changing spatial and temporal domains still present obstacles. The main reasons for this are the dynamic nature of a scene and the difficulty of representing and incorporating the feature space of objects of varying sizes into a prediction model. To overcome the aforementioned issues, this paper proposes a parallel multi-size receptive field units framework that leverages the majority of the CNN layer’s features, allowing for the representation and participation in the model prediction of the features of objects of all sizes. The proposed method utilizes features generated from lower to higher layers. As a result, different object scales can be handled at different framework depths, and various environmental densities can be estimated. However, the inclusion of the vast majority of layer features in the prediction model has a number of negative effects on the prediction’s outcome. Asymmetric non-local attention and the channel weighting module of a feature map are proposed to handle noise and background details and re-weight each channel to make it more sensitive to important features while ignoring irrelevant ones, respectively. While the output predictions of some layers have high bias and low variance, those of other layers have low bias and high variance. Using stack ensemble meta-learning, we combine individual predictions made with lower-layer features and higher-layer features to improve prediction while balancing the tradeoff between bias and variance. The UCF CC 50 dataset and the ShanghaiTech dataset have both been subjected to extensive testing. The results of the experiments indicate that the proposed method is effective for dense distributions and objects of various sizes.

Dark matter and motion of bodies in space

This paper investigates the degree of influence of the gravitational field of dark matter on the laws of motion of bodies in a medium in a restricted two-body problem, when a test body (planet, asteroid, artificial satellite of a star, in particular, the Sun, etc.) has its own rotation, i. e. own angular momentum impulse. The study was carried out within the framework of the post-Newtonian approximation of the general theory of relativity. In accordance with the latest experimental data, hypotheses about the average densities of dark matter ρD.M. and visible matter ρvis. in planetary systems are accepted. In particular, in the Solar system the following is accepted: ρD.M ≈ 2,8 · 10–19 g · cm–3, ρvis ≈ 3 · 10–20 g · cm–3 and ρΣ = ρvis + ρD.M ≈ 3,1 · 10–19 g · cm–3. In the post-Newtonian approximation of the general theory of relativity, the equation for the trajectory of a rotating test body with respect to ρΣ is derived, and working formulas are obtained that give the laws of secular changes in the direction of the vector of the proper angular momentum impulse of the test body and the modulus of this vector. It is shown that accounting ρD.M changes the magnitude of the periastron shift. For example, in the Solar System when taking into account ρvis, all the planets except Pluto have a directly shifted perihelion in the post-Newtonian approximation of the general theory of relativity. When taking into account ρΣ the planets from Mercury to Saturn included, they have a direct shift of perihelion, and Uranus, Neptune, Pluto have the reverse (against the planets in orbit). There is also a secular change in the eccentricity of the orbit. The formula is derived that can be used to calculate the secular deviation of the translational motion of a rotating body from motion in a plane. Accounting ρΣ enhances deviation. It is emphasized that all the noted effects for planetary systems in the vicinity of neutron stars, radio pulsars and other dense objects can be many orders of magnitude greater than in the solar system.

Numerical refractive index correction for the stitching procedure in tomographic quantitative phase imaging.

Tomographic quantitative phase imaging (QPI) lacks an absolute refractive index value baseline, which poses a problem when large dense objects extending over multiple fields of view are measured volume by volume and stitched together. Some of the measurements lack the natural baseline value that is provided by the mounting medium with a known refractive index. In this work, we discuss the problem of the refractive index (RI) baseline of individual reconstructed volumes that are deprived of access to mounting medium due to the extent of the object. The solution of this problem is provided by establishing the RI offsets based on the overlapping regions. We have proven that the process of finding the offset RI values may be justifiably reduced to the analogous procedure in the 2D baseline correction (2D-BC). Finally, we proposed the enhancement of the state-of-the-art 2D-BC procedure previously introduced in the context of 2D QPI. The processing is validated at the examples of a synthetic dataset and a liver organoid.

Towards Efficient Detection for Small Objects via Attention-Guided Detection Network and Data Augmentation.

Small object detection has always been a difficult direction in the field of object detection, especially the detection of small objects in UAV aerial images. The images captured by UAVs have the characteristics of small objects and dense objects. In order to solve these two problems, this paper improves the performance of object detection from the aspects of data and network structure. In terms of data, the data augmentation strategy and image pyramid mechanism are mainly used. The data augmentation strategy adopts the method of image division, which can greatly increase the number of small objects, making it easier for the algorithm to be fully trained during the training process. Since the object is denser, the image pyramid mechanism is used. During the training process, the divided images are up-sampled into three different sizes, and then sent to three different detectors respectively. Finally, the detection results of the three detectors are fused to obtain the final detection results. The small object itself has few pixels and few features. In order to improve the detection performance, it is necessary to use context. This paper adds attention mechanism to the yolov5 network structure, while adding a detection head to the underlying feature map to make the network structure pay more attention to small objects. By using data augmentation and improved network structure, the detection performance of small objects can be significantly improved. The experiment in this paper is carried out on the Visdrone2019 dataset and DOTA dataset. Through experimental verification, our proposed method can significantly improve the performance of small object detection.

Multi-Object Detection Method in Construction Machinery Swarm Operations Based on the Improved YOLOv4 Model.

To handle the problem of low detection accuracy and missed detection caused by dense detection objects, overlapping, and occlusions in the scenario of complex construction machinery swarm operations, this paper proposes a multi-object detection method based on the improved YOLOv4 model. Firstly, the K-means algorithm is used to initialize the anchor boxes to improve the learning efficiency of the depth features of construction machinery objects. Then, the pooling operation is replaced with dilated convolution to solve the problem that the pooling layer reduces the resolution of feature maps and causes a high missed detection rate. Finally, focus loss is introduced to optimize the loss function of YOLOv4 to improve the imbalance of positive and negative samples during the model training process. To verify the effectiveness of the above optimizations, the proposed method is verified on the Pytorch platform with a self-build dataset. The experimental results show that the mean average precision(mAP) of the improved YOLOv4 model for multi-object detection of construction machinery can reach 97.03%, which is 2.16% higher than that of the original YOLOv4 detection network. Meanwhile, the detection speed is 31.11 fps, and it is reduced by only 0.59 fps, still meeting the real-time requirements. The research lays a foundation for environment perception of construction machinery swarm operations and promotes the unmanned and intelligent development of construction machinery swarm operations.

FEA-Swin: Foreground Enhancement Attention Swin Transformer Network for Accurate UAV-Based Dense Object Detection.

UAV-based object detection has recently attracted a lot of attention due to its diverse applications. Most of the existing convolution neural network based object detection models can perform well in common object detection cases. However, due to the fact that objects in UAV images are spatially distributed in a very dense manner, these methods have limited performance for UAV-based object detection. In this paper, we propose a novel transformer-based object detection model to improve the accuracy of object detection in UAV images. To detect dense objects competently, an advanced foreground enhancement attention Swin Transformer (FEA-Swin) framework is designed by integrating context information into the original backbone of a Swin Transformer. Moreover, to avoid the loss of information of small objects, an improved weighted bidirectional feature pyramid network (BiFPN) is presented by designing the skip connection operation. The proposed method aggregates feature maps from four stages and keeps abundant information of small objects. Specifically, to balance the detection accuracy and efficiency, we introduce an efficient neck of the BiFPN network by removing a redundant network layer. Experimental results on both public datasets and a self-made dataset demonstrate the performance of our method compared to the state-of-the-art methods in terms of detection accuracy.

Research on Fingerprint Security Based on Improved Yolo Algorithm

The quantitative identification technology based on the statistical law of fingerprint features has become a new research difficulty and focus, and the automatic detection and classification of fingerprint features are the basis for realizing automatic fingerprint feature statistics. In this paper, a YOLO-based fingerprint feature detection method was proposed. First, a fingerprint feature dataset was established, which contained a total of 4,000 annotated fingerprint images; then, according to the characteristics of small size and dense distribution of fingerprint feature points, the YOLO network structure was improved, the original large target feature detection layer by 32-fold downsampling was deleted, and a new small feature fusion layer was added; the FPN, PAN, and SPP structures were used to achieve local and global feature extraction through multiple-scale fusion methods; finally, the SE channel attention mechanism module was added to effectively enhance the model robustness and detection ability of dense small objects. The experimental results show that compared with the improved FP-YOLO model of the original model, when the detection speed is basically unchanged, the mAP0.5 value is increased from 93.0% to 97.4%, and the weight is reduced by 3/4.

YOLOX-SAR: High-Precision Object Detection System Based on Visible and Infrared Sensors for SAR Remote Sensing

The object detection technology for Synthetic Aperture Radar (SAR) image generation is of great significance in signal processing, radar imaging and other fields. SAR image is the image data obtained from the electromagnetic wave echo from the radar to the object after processing the range and the azimuth respectively. However, the sizes of the objects to be detected change dramatically and the detection difficulty increases, because the scattering characteristics of electromagnetic waves have a great influence on SAR images. At the same time, the large areas of background information will contain confused geographical elements, which seriously affect the detection performance. Therefore, we propose the YOLOX-SAR algorithm to solve these problems. Based on YOLOX, we combine the adaptive activation function Meta-ACON with the SPP module in Backbone to improve the feature extraction ability. We also integrate the Convolutional Block Attention Model (CBAM) at FPN to find the attention area in scenes with dense objects. Furthermore, many useful strategies, such as data enhancement and MS-testing, are provided to improve our proposed YOLOX-SAR. Apart from that, in order to objectively demonstrate the competitive performance of YOLOX-SAR, we compare it with other object detection algorithms on the sensor acquisition dataset of NWPU VHR-10. There remain a large amount of research results demonstrating that YOLOX-SAR possesses good performance as well as interpretation ability in SAR image object detection. The mAP of YOLOX-SAR reaches 89.56%, which is 4.04% higher than that of the previous YOLOX algorithm. YOLOX-SAR, as an improved version of YOLO series, can still be detected in real-time and its FPS reaches 67, keeping the characteristic of fast detection speed of the YOLOX series. Experimental results and analysis show that under the complex electromagnetic scattering background, YOLOX-SAR can ensure the authenticity and effectiveness of the dataset collected by the sensors. Meanwhile, YOLOX-SAR ensures a good balance between the accuracy and running time of object detection. Therefore, the algorithm has good detection performance for objects of various sizes in SAR images.

A machine learning algorithm for direct detection of axion-like particle domain walls

The Global Network of Optical Magnetometers for Exotic physics searches (GNOME) conducts an experimental search for certain forms of dark matter based on their spatiotemporal signatures imprinted on a global array of synchronized atomic magnetometers. The experiment described here looks for a gradient coupling of axion-like particles (ALPs) with proton spins as a signature of locally dense dark matter objects such as domain walls. In this work, stochastic optimization with machine learning is proposed for use in a search for ALP domain walls based on GNOME data. The validity and reliability of this method were verified using binary classification. The projected sensitivity of this new analysis method for ALP domain-wall crossing events is presented.

R-YOLO: A YOLO-Based Method for Arbitrary-Oriented Target Detection in High-Resolution Remote Sensing Images.

In view of the existence of remote sensing images with large variations in spatial resolution, small and dense objects, and the inability to determine the direction of motion, all these components make object detection from remote sensing images very challenging. In this paper, we propose a single-stage detection network based on YOLOv5. This method introduces the MS Transformer module at the end of the feature extraction network of the original network to enhance the feature extraction capability of the network model and integrates the Convolutional Block Attention Model (CBAM) to find the attention area in dense scenes. In addition, the YOLOv5 target detection network is improved by incorporating a rotation angle approach from the a priori frame design and the bounding box regression formulation to make it suitable for rotating frame-based detection scenarios. Finally, the weighted combination of the two difficult sample mining methods is used to improve the focal loss function, so as to improve the detection accuracy. The average accuracy of the test results of the improved algorithm on the DOTA data set is 77.01%, which is higher than the previous detection algorithm. Compared with the average detection accuracy of YOLOv5, the average detection accuracy is improved by 8.83%. The experimental results show that the algorithm has higher detection accuracy than other algorithms in remote sensing scenes.

Semantic-Edge-Supervised Single-Stage Detector for Oriented Object Detection in Remote Sensing Imagery

In recent years, significant progress has been made in arbitrary-oriented object detection. Different from natural images, object detection in aerial images remains its problems and challenges. Current feature enhancement strategies in this field mainly focus on enhancing the local critical response of the target while ignoring the target’s contextual information, which is indispensable for detecting remote sensing targets in complex backgrounds. In this paper, we innovatively combine semantic edge detection with arbitrary-oriented object detection and propose a feature enhancement network base on a semantic edge supervision module (SES) that realizes an attention-like mechanism in three dimensions of space, channel, and pyramid level. It helps the network pay attention to the edge features of targets at multiple scales to obtain more regression clues. Furthermore, to solve the problem of dense objects with different directions in remote sensing images, we propose a rotation-invariant spatial pooling pyramid (RISPP) to extract the features of objects from multiple orientations. Based on the two feature enhancement modules, we named the network SE2-Det; extensive experiments on large public datasets of aerial images (DOTA and UCAS-AOD) validate our approach’s effectiveness and demonstrate our detector’s superior performance.

An Application of Pixel Interval Down-Sampling (PID) for Dense Tiny Microorganism Counting on Environmental Microorganism Images

This paper proposes a novel pixel interval down-sampling network (PID-Net) for dense tiny object (yeast cells) counting tasks with higher accuracy. The PID-Net is an end-to-end convolutional neural network (CNN) model with an encoder–decoder architecture. The pixel interval down-sampling operations are concatenated with max-pooling operations to combine the sparse and dense features. This addresses the limitation of contour conglutination of dense objects while counting. The evaluation was conducted using classical segmentation metrics (the Dice, Jaccard and Hausdorff distance) as well as counting metrics. The experimental results show that the proposed PID-Net had the best performance and potential for dense tiny object counting tasks, which achieved 96.97% counting accuracy on the dataset with 2448 yeast cell images. By comparing with the state-of-the-art approaches, such as Attention U-Net, Swin U-Net and Trans U-Net, the proposed PID-Net can segment dense tiny objects with clearer boundaries and fewer incorrect debris, which shows the great potential of PID-Net in the task of accurate counting.

Swin-Transformer-Enabled YOLOv5 with Attention Mechanism for Small Object Detection on Satellite Images

Object detection has made tremendous progress in natural images over the last decade. However, the results are hardly satisfactory when the natural image object detection algorithm is directly applied to satellite images. This is due to the intrinsic differences in the scale and orientation of objects generated by the bird’s-eye perspective of satellite photographs. Moreover, the background of satellite images is complex and the object area is small; as a result, small objects tend to be missing due to the challenge of feature extraction. Dense objects overlap and occlusion also affects the detection performance. Although the self-attention mechanism was introduced to detect small objects, the computational complexity increased with the image’s resolution. We modified the general one-stage detector YOLOv5 to adapt the satellite images to resolve the above problems. First, new feature fusion layers and a prediction head are added from the shallow layer for small object detection for the first time because it can maximally preserve the feature information. Second, the original convolutional prediction heads are replaced with Swin Transformer Prediction Heads (SPHs) for the first time. SPH represents an advanced self-attention mechanism whose shifted window design can reduce the computational complexity to linearity. Finally, Normalization-based Attention Modules (NAMs) are integrated into YOLOv5 to improve attention performance in a normalized way. The improved YOLOv5 is termed SPH-YOLOv5. It is evaluated on the NWPU-VHR10 dataset and DOTA dataset, which are widely used for satellite image object detection evaluations. Compared with the basal YOLOv5, SPH-YOLOv5 improves the mean Average Precision (mAP) by 0.071 on the DOTA dataset.

Moving Object Tracking in Satellite Videos by Kernelized Correlation Filter Based on Color-Name Features and Kalman Prediction

This paper studies moving object tracking in satellite videos. For the satellite videos, the object size in the images may be small, the object may be partly occluded, and the image may contain an area resembling dense objects. To handle the above problems, this paper puts forward a kernelized correlation filter based on the color-name feature and Kalman prediction. The original image is mapped to the color-name feature space so that the tracker can process the image with multichannel color features. The Kalman filter is used to predict the moving object position in the tracking process, and the detection area is determined according to the predicted position. The Kalman filter is updated with the detection results to improve the tracking accuracy. The proposed algorithm is tested on Jilin-1 datasets. Compared with the other seven tracking algorithms, the experiment results show that the proposed algorithm has stronger robustness for several complex situations such as rapid target motion and similar object interference. Besides, it is also shown that the proposed algorithm can prevent the problem of tracking failure when the moving object is partially occluded.

Gravitational waves from fully general relativistic oscillon preheating

As long-lived quasisolitons from the fragmentation of a scalar condensate, oscillons may dominate the preheating era after inflation. During this period, stochastic gravitational waves can also be generated. We quantify the gravitational-wave production in this period with simulations accounting for full general relativity to capture all possible nonperturbative effects. We compute the gravitational-wave spectra across a range of choices of the oscillon preheating models and compare our results to a conventional perturbative approach on a Friedmann-Lema\^{\i}tre-Robertson-Walker (FLRW) background. We clarify the gauge ambiguities in computing induced gravitational waves from scenarios where dense nonperturbative objects such as oscillons are being formed. In particular, we find that the synchronous gauge tends to contain large artificial enhancements in the gravitational-wave spectrum due to gauge modes if gravity plays an important role in the formation of the oscillons, while other gauge choices, such as the radiation gauge or a suitably chosen ``1+log'' gauge, can efficiently reduce the contributions of gauge modes. The full general-relativistic simulations indicate that gravitational-wave spectra obtained from the perturbative approach on the FLRW background are fairly accurate, except when oscillon formation induces strong gravitational effects, for which case there can be an order unity enhancement.