Missed Detection Rate Research Articles

ABYOLOv4: improved YOLOv4 human object detection based on enhanced multi-scale feature fusion

The purpose of human object detection is to obtain the number of people and their position in images, which is one of the core problems in the field of machine vision. However, the high missing detection rate from small- and medium-sized human bodies due to the large variety of human scale in human object detection tasks still influences the performance of human object detection. To solve the above problem, this paper proposed an improved ASPP_BiFPN_YOLOv4 (ABYOLOv4) method to detect human object detection. In detail, Atrous Spatial Pyramid Pooling (ASPP) module was used to replace the original Spatial Pyramid Pooling module to increase the receptive field level of the network and improve the perception ability of multi-scale targets. Then, the original Path Aggregation Network (PANet) multi-scale fusion module was replaced by the self-built bi-layer bidirectional feature pyramid network (Bi-FPN). Meanwhile, a new feature was imported into the proposed model to reuse the mid- and low-level features, which could enhance the ability of the network to express the characteristics of small- and medium-sized targets. Finally, the standard convolution in Bi-FPN was replaced by depth-separable convolution to make the network achieve the balance of accuracy and the number of parameters. To identify the performance of the proposed ABYOLOv4 model, the human object detection experiment is carried out by using the public data set of VOC2007 and VOC2012, the improved YOLOv4 algorithm is 0.5% higher than the original AP algorithm, and the weight file size of the model is reduced by 45.3 M. The experimental results demonstrated that the proposed ABYOLOv4 network has higher accuracy and lower computational cost for human target detection.

Multi visual images fusion approach for metro tunnel defects based on saliency optimization of pixel level defect image features

The multi-vision defect sensing system, lining composed primarily of IRT and RGB cameras, allows for automatic identification and extraction of small surface ailments, greatly enhancing detection efficiency. However, the presence of various issues like train vibration, inconsistent lighting, fluctuations in temperature and humidity leads to the images showing inadequate uniformity in illumination, blurriness, and a decrease in the level of detail. The above issues have led to unsatisfactory fusion processing results for multiple visual images and increased missed detection rates. To address the above-mentioned issue, multi visual images fusion approach for subway tunnel defects based on saliency optimization of pixel level defect image features is proposed. The approach initially analyses the train’s motion status and image blurring conditions. It then eliminates the dynamic blurring in the image. Secondly, Image weights are allocated based on the uniformity of visible light image illumination in the tunnel, as well as real-time temperature and humidity. Finally, image feature extraction and fusion are performed by a U-Net network that integrates channel attention mechanisms. The entire experiment was carried out on a dataset consisting of leakage data from the tunnel lining of Shanghai Metro and tunnel defect data from Beijing Metro. The experimental results demonstrate that this approach improves the image pixel value variation rate by 39.7%, enhances the edge quality by 23%, and outperforms similar approach in terms of average gradient, gradient quality, and sum of difference correlation with improvements of 15.9%, 7.3%, and 26.6% respectively.

Optical flow-based vertical angular rate fault detection on UAVs

This paper discusses optical flow-based vertical angular rate fault detection based-on real fight IMU data and camera image sequences. The fight test platform utilizes a gimbal stabilized downward looking camera that is why only the angular rate around the inertial vertical axis can be estimated. Improvements relative to the previous work of the authors are the application of forward-backward optical flow to filter outliers, the direct detection of angular rate faults without the need for Euler angles or GPS velocities and tuning and verification on real images and fight data. After presenting the optical flow equations and the method for image-based vertical angular rate estimation the fight test scenarios and video processing steps are introduced. Four different Artificial errors are added to the IMU measurements and the fault detection applies residual thresholding and up-down counters. Tuning is based on false alarm and missed detection rates and the application of receiver operating characteristics. Finally, the estimation results and possible further improvements are presented.

DSNet: A Computer Vision‐Based Detection and Corrosion Segmentation Network for Corroded Bolt Detection in Tunnel

Corroded bolt detection has been confirmed as a major issue in the structure health monitoring (SHM) of tunnels. However, detection‐only methods will miss the corroded bolts, arising from the small rust area. In this study, the task is divided ingeniously into two parallel tasks, i.e., bolt detection and pixel‐level rust segmentation, and the objective is fulfilled by taking the intersection of the two tasks, with the aim of enhancing the performance. To be specific, a detection and segmentation network (DSNet) is proposed based on multitask learning, leading to reduced false and missed detection rates. The coordinate attention module enhancing the focus of bolts in tunnel patches is incorporated in the detection branch, and the cross‐stage partial‐based decoder which can more accurately determine whether a pixel pertains to the corrosion area is employed in the segmentation branch. The mentioned branches share the same backbone to simplify the model. Sufficient comparisons and ablation experiments are performed to prove the superiority of the proposed algorithm based on the corroded bolt dataset captured from a real subway tunnel, which is publicly available in https://github.com/StreamHXX/Tunnel-lining-disease-image.

Robust Data Association Against Detection Deficiency for Semantic SLAM

Robust and accurate object association is essential for precise 3D object landmark inference in semantic Simultaneous Localization and Mapping (SLAM), and yet remains challenging due to the detection deficiency caused by high miss detection rate, false alarm, occlusion and limited field-of-view, etc. The 2D location of an object is a crucial complementary cue to the appearance feature, especially in the case of associating objects across frames under large viewpoint changes. However, motion model or trajectory pattern based methods struggle to infer object motion reliably with a moving camera. In this paper, by exploiting the local projective warping consistency, a local homography based 2D motion inference method is proposed to sequentially estimate the object location along with uncertainty. By integrating the deep appearance feature and semantic information, an object association method, named HOA, which is robust to detection deficiency is proposed. Experimental evaluations suggest that the proposed motion prediction method is capable of maintaining a low cumulative error over a long duration, which enhances the object association performance in both accuracy and robustness. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This work aims to consistently associate 2D detection boxes corresponding to the same 3D object across images. In tasks of landmark-based navigation, collision avoidance, grasping and manipulation, objects in the task space are commonly simplified into 3D enveloping surfaces (e.g. cuboid or ellipsoid) by using 2D object detection boxes from multiple image views, and accurate data association is a prerequisite for precise enveloping surface reconstruction. This problem remains challenging considering the imperfect object detections, the appearance similarity of objects and the unpredictable trajectory of the moving camera. This work proposes a long-term reliable 2D location prediction algorithm that is capable of handling the complex motion of the target. Along with the appearance feature extracted by a retrain-free deep learning based model, this work proposes an object association method that can simultaneously deal with multiple objects with unknown object categories under the moving camera scenario.

A Dual-Tree–Complex Wavelet Transform-Based Infrared and Visible Image Fusion Technique and Its Application in Tunnel Crack Detection

Computer vision methods have been widely used in recent years for the detection of structural cracks. To address the issues of poor image quality and the inadequate performance of semantic segmentation networks under low-light conditions in tunnels, in this paper, infrared images are used, and a preprocessing method based on image fusion technology is developed. First, the DAISY descriptor and the perspective transform are applied for image alignment. Then, the source image is decomposed into high- and low-frequency components of different scales and directions using DT-CWT, and high- and low-frequency subband fusion rules are designed according to the characteristics of infrared and visible images. Finally, a fused image is reconstructed from the processed coefficients, and the fusion results are evaluated using the improved semantic segmentation network. The results show that using the proposed fusion method to preprocess images leads to a low false alarm rate and low missed detection rate in comparison to those using the source image directly or using the classical fusion algorithm.

Automatic GNSS Ionospheric Scintillation Detection with Radio Occultation Data Using Machine Learning Algorithm

Ionospheric scintillation often occurs in the polar and equator regions, and it can affect the signals of the Global Navigation Satellite System (GNSS). Therefore, the ionospheric scintillation detection applied to the polar and equator regions is of vital importance for improving the performance of satellite navigation. GNSS radio occultation is a remote sensing technique that primarily utilizes GNSS signals to study the Earth’s atmosphere, but its measurement results are susceptible to the effects of ionospheric scintillation. In this study, we propose an ionospheric scintillation detection algorithm based on the Sparrow-Search-Algorithm-optimized Extreme Gradient Boosting model (SSA-XGBoost), which uses power spectral densities of the raw signal intensities from GNSS occultation data as input features to train the algorithm model. To assess the performance of the proposed algorithm, we compare it with other machine learning algorithms such as XGBoost and a Support Vector Machine (SVM) using historical ionospheric scintillation data. The results show that the SSA-XGBoost method performs much better compared to the SVM and XGBoost models, with an overall accuracy of 97.8% in classifying scintillation events and a miss detection rate of only 12.9% for scintillation events with an unbalanced GNSS RO dataset. This paper can provide valuable insights for designing more robust GNSS receivers.

Research on Transformer Omnidirectional Partial Discharge Ultrasound Sensing Method Combining F-P Cavity and FBG.

To achieve omnidirectional sensitive detection of partial discharge (PD) in transformers and to avoid missing PD signals, a fiber optic omnidirectional sensing method for PD in transformers combined with the fiber Bragg grating (FBG) and Fabry-Perot (F-P) cavity is proposed. The fiber optic omnidirectional sensor for PD as a triangular prism was developed. The hollow structure of the probe was used to insert a single-mode fiber to form an F-P cavity. In addition, the three sides of the probe were used to form a diaphragm-type FBG sensing structure. The ultrasound sensitization diaphragm was designed based on the frequency characteristics of PD in the transformer and the vibration model of the diaphragm in the liquid environment. The fiber optic sensing system for PD was built and the performance test was conducted. The results show that the resonant frequency of the FBG acoustic diaphragm is around 20 kHz and that of the F-P cavity acoustic diaphragm is 94 kHz. The sensitivity of the developed fiber optic sensor is higher than that of the piezoelectric transducer (PZT). The lower limit of PD detection is 68.72 pC for the FBG sensing part and 47.97 pC for the F-P cavity sensing part. The directional testing of the sensor and its testing within a transformer simulation model indicate that the proposed sensor achieves higher detection sensitivity of PD in all directions. The omnidirectional partial discharge ultrasound sensing method proposed in this paper is expected to reduce the missed detection rate of PD.

A Threshold Design Method of Aircraft Sensor Fault Detection Monitor

A design method is proposed based on probability and statistics to solve the issue that the contradiction between false alarm rate and missed detection rate of sensor comparing monitor threshold definition. The sensor tracking error probability density function is deduced based on the probability, the false alarm rate and missed detection rate are calculated based on the different monitor threshold. Then a criterion is provided in according with sensor failure rate, aircraft dispatch reliability and monitor failure rate. In order to verify it, a quad redundant rate gyro used on aircraft is analysed and an advanced comparing monitor threshold is calculated. Comparing with the original threshold, the monitor coverage rate with the new threshold is increased 1.42 times than the original one. The actual flight date is researched to verify that the monitor failure rate reach to the minimum value 3.91*10-07/Fh which decrease one order of magnitude comparing with the original one 3.5*10-06/Fh. It could be verified that this method could achieve the balance between the safety and dispatch reliability requirement.

The possibility of rainfall nowcasting using atmospheric electric field

The possibility of rainfall nowcasting using a new parameter — atmospheric electric field (potential gradient, PG) — is investigated. The temporal behavior of the PG measured by a ground-based electric field mill before rainfall was analyzed at the University of Reading Atmospheric Observatory (51.441°N, 0.938°W, 66 m) in 2007–2021 and the Nor Amberd research station (40.367°N, 45.258°E, 2000 m) in 2013–2018. The PG tend to show abnormal oscillations before the precipitation cloud arrived at the PG measuring site. Several rules were applied to determine rainfall and PG oscillation events. Statistical analysis was conducted, and the results showed that PG oscillations occurred before and overlap with 48% (54%) of rainfall events, and PG oscillations occurred in one hour before 68% (77%) of rainfall events started at the University of Reading Atmospheric Observatory site (Nor Amberd research station site). A PG-based rainfall nowcasting model is proposed, which is constructed using machine learning techniques Fully Connected Neural Network and Tree-based Pipeline Optimization Tool. The hit rate, false alarm rate, missed detection rate, TS score of the model reached 93.8% (94.6%), 36% (37.4%), 37.7% (39.3%), 46.1% (44.5%) respectively, which is comparable to that of the state-of-the-art rainfall nowcasting methods. And the heavy rainfall hit rate reached 41.5% (36.4%). PG show a great potential to be used as a new parameter for rainfall nowcasting. And the effect of rainfall nowcasting method that proposed in this paper is barely affected by the weather conditions and the PG measuring range of different sites. Moreover, PG stands out among many other parameters because long-term and stable PG measurements can be obtained at a very low cost using an electric field mill.

A health status estimation method based on interpretable neural network observer for HVs

Estimating the health status is a crucial step in learning about the health of hypersonic vehicles beforehand. The estimation results can be used to detect abnormal states and provide data reference for fault diagnosis. However, certain conventional neural network-based estimate techniques rely heavily on data and have limited model interpretability, which challenges the accuracy of the estimation results. This research aims to address the problems of data dependency and model interpretability in estimation models. In this study, a block interpretable neural network model with constraints on the trajectory and attitude equations is established. On the basis of the interpretable neural network model, two health status estimation methods are proposed: one that is unsupervised and the other that is supervised. Additionally, in the supervised health status estimate approach, an FC-LN-Mish structure is created to fit the relationship between the fault residual and the fault state parameters. The results indicate that the proposed estimation methods can fit the system mechanism relationship more accurately, improve the model interpretability, reduce data dependency, and ensure high estimation efficiency and precision. The FC-LN-Mish structure can reduce the missed detection rate and false detection rate to some extent, and perform better than other models under the low fault deviation degree. In conclusion, the interpretable neural network model-based observers accurately observe the health status parameters of rudders and RCS, reduce data dependence and data processing costs, and have better performance under high uncertainty interference. It provides effective method for online health estimation.

End-to-end Remote Sensing Image Aircraft Target Detection and Fine-grained Recognition Framework

In view of the practical application of target detection and recognition tasks for remote sensing, an end-to-end aircraft target detection and fine-grained recognition framework is proposed. It can accurately and quickly implement detection and recognition in an end-to-end way. The main network of the framework adopts the design ideas of target detection and fine-grained recognition methods using candidate region extract and visual attention, making sure the accuracy of detection and recognition. Then, to solve the problem of high false detection rate and missed detection rate of densely arranged targets, we propose the re-detection mechanism. To minimize the large amounts of calculations of deep networks and improve real-time performance, we introduce depthwise separable convolution to optimize networks. Finally, a weight mapping idea based on transfer learning is adopted to solve the problem of data labeling and also helps the detection and fine-grained recognition. The results prove that the proposed framework has good robustness, versatility, and efficiency in aircraft detection and fine-grained recognition tasks.

Enhanced Detection Method for Small and Occluded Targets in Large-Scene Synthetic Aperture Radar Images

Ship detection in large-scene offshore synthetic aperture radar (SAR) images is crucial in civil and military fields, such as maritime management and wartime reconnaissance. However, the problems of low detection rates, high false alarm rates, and high missed detection rates of offshore ship targets in large-scene SAR images are due to the occlusion of objects or mutual occlusion among targets, especially for small ship targets. To solve this problem, this study proposes a target detection model (TAC_CSAC_Net) that incorporates a multi-attention mechanism for detecting marine vessels in large-scene SAR images. Experiments were conducted on two public datasets, the SAR-Ship-Dataset and high-resolution SAR image dataset (HRSID), with multiple scenes and multiple sizes, and the results showed that the proposed TAC_CSAC_Net model achieves good performance for both small and occluded target detection. Experiments were conducted on a real large-scene dataset, LS-SSDD, to obtain the detection results of subgraphs of the same scene. Quantitative comparisons were made with classical and recently developed deep learning models, and the experiments demonstrated that the proposed model outperformed other models for large-scene SAR image target detection.

Feature extraction and segmentation of pavement distress using an improved hybrid task cascade network

ABSTRACT An improved Hybrid Task Cascade (HTC) instance segmentation model was proposed to segment pavement surface distress intelligently and accurately. First, the automatic color equalization (ACE) algorithm was performed to preprocess the dataset. Then, HTC and the proposed HTC were described. The proposed HTC was modified using the ResNeXt101 network, balanced feature pyramid (BFP) and deformable convolution (DCN). The Soft non-maximum suppression method (Soft-NMS) and balanced L1 loss were used to supplement the above work. Finally, the average detection precision (AP) and the segmentation AP indexes were assessed by comparison and contrast. Experimental results demonstrated that the original HTC has crack miss detection and false detection problems. In contrast, the improved HTC has a recall rate as high as 96.1%, indicating a low miss-detection rate. Compared with other mainstream instance segmentation models, the improved HTC still leads in various indicators on our or public datasets. Comparisons between the predicted and true distress areas were analysed, the relative error of which was reasonable and satisfactory. The results help judge the degree of pavement distress and provide data support for repair and maintenance, which has essential engineering significance.

Online micro defects detection for ductile cast iron pipes based on twin light photometric stereo

Ductile cast iron pipes (DCIPs) are widely utilized in water and gas supply due to their exceptional mechanical and material properties. Quality control during production is paramount for optimal performance, with surface defect detection being of utmost importance. Currently, production sites primarily rely on manual visual inspection, which presents several challenges, including high labor intensity, low efficiency, and subjectivity in defect judgment. 2D detection algorithms based on single images struggle to effectively detect small defects such as pores, pinholes, and scratches, resulting in high missed detection rates and vulnerability to interference from oil and water stains. To address these issues, this paper proposes a symmetrical twin-light photometric stereo (TLPS) detection algorithm based on photometric stereo theory, and image correction algorithm for mitigating the effects of ambient light. A visual inspection prototype system was designed and successfully implemented at a production site. The system comprises eight visual inspection units, each equipped with a high-speed line scan camera and two bar light sources for image acquisition of moving DCIP. Under the Lambertian model and parallel light source assumption, the gradient expression in the texture direction of the DCIP surface is derived based on light source symmetry. Production experiment results demonstrate that the gradient map obtained through the proposed TLPS effectively enhances defect contrast and accurately detects and locates micro defects such as pores, pinholes, and scratches on cast pipe surfaces. This verifies the effectiveness of the TLPS algorithm, addresses the shortcomings of traditional 2D detection algorithms, and lays the foundation for subsequent online detection of DCIP surface defects.

Exploring Deep Learning for Adaptive Energy Detection Threshold Determination: A Multistage Approach

The concept of spectrum sensing has emerged as a fundamental solution to address the growing demand for accessing the limited resources of wireless communications networks. This paper introduces a straightforward yet efficient approach that incorporates multiple stages that are based on deep learning (DL) techniques to mitigate Radio Frequency (RF) impairments and estimate the transmitted signal using the time domain representation of received signal samples. The proposed method involves calculating the energies of the estimated transmitted signal samples and received signal samples and estimating the energy of the noise using these estimates. Subsequently, the received signal energy and the estimated noise energy, adjusted by a correction factor (k), are employed in binary hypothesis testing to determine the occupancy of the wireless channel under investigation. The proposed system demonstrates encouraging outcomes by effectively mitigating RF impairments, such as carrier frequency offset (CFO), phase offset, and additive white Gaussian noise (AWGN), to a considerable degree. As a result, it enables accurate estimation of the transmitted signal from the received signal, with 3.85% false alarm and 3.06% missed detection rates, underscoring the system’s capability to adaptively determine a decision threshold for energy detection.

An Edge-Guided Deep Learning Solar Panel Hotspot Thermal Image Segmentation Algorithm

To overcome the deficiencies in segmenting hot spots from thermal infrared images, such as difficulty extracting the edge features, low accuracy, and a high missed detection rate, an improved Mask R-CNN photovoltaic hot spot thermal image segmentation algorithm has been proposed in this paper. Firstly, the edge image features of hot spots were extracted based on residual neural networks. Secondly, by combining the feature pyramid structure, an edge-guided feature pyramid structure was designed, and the hot spot edge features were injected into a Mask R-CNN network. Thirdly, an infrared spatial attention module was introduced into the Mask R-CNN network when feature extraction and the infrared features of the detected hot spots were enhanced. Fourthly, the size ratio of the candidate frames was adjusted self-adaptively according to the structural characteristics of the aspect ratio of the hot spots. Finally, the validation experiments were conducted, and the results demonstrated that the hot spot contours of thermal infrared images were enhanced through the algorithm proposed in this paper, and the segmentation accuracy was significantly improved.

A High-Confidence Intelligent Measurement Method for Aero-Engine Oil Debris Based on Improved Variational Mode Decomposition Denoising

This paper presents an effective method for measuring oil debris with high confidence to ensure the wear monitoring of aero-engines, which suffers from severe noise interference, weak signal characteristics, and false detection. First, an improved variational mode decomposition algorithm is proposed, which combines wavelet transform and interval threshold processing to suppress the complex noise interference on the signal. Then, a long-short-term memory neural network with deep scattering spectrum preprocessing is used to identify the signal characteristics under the multi-resolution analysis framework. The optimal hyperparameters are automatically configured using Bayesian optimization to solve the problem of weak, distorted, and hard-to-extract signal characteristics. Finally, a detection algorithm based on multi-window fusion judgment is applied to improve the confidence of the detection process, reduce the false detection and false alarm rate, and calculate the debris size information according to the sensor principle. The experimental results show that the proposed method can extract debris signals from noise with a signal-to-noise ratio improvement of more than 9 dB, achieve a high recognition accuracy of 99.76% with a missed detection rate of 0.24%, and output size information of debris to meet the need for aero-engine oil debris measurement.

Utilize DBN and DBSCAN to detect selective forwarding attacks in event-driven wireless sensors networks

Malicious nodes launching selective forwarding attacks jeopardize network reliability in event-driven wireless sensor networks. Swift detection and exclusion of these nodes are vital, especially in harsh environments where normal nodes also suffer from decreased forwarding rates due to poor channel conditions. To solve this problem, this paper proposes a deep belief network (DBN) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) based detection scheme against selective forwarding attacks under harsh environments. The DBN algorithm extracts and analyzes behavior features of nodes, while the DBSCAN clustering algorithm effectively distinguishes between malicious and normal nodes during selective forwarding attacks. Simulation results demonstrate the scheme’s effectiveness with a missed detection rate (MDR) of approximately 2% and a false detection rate (FDR) below 5% in harsh environments, providing a robust solution for ensuring the integrity and efficiency of data transmission in event-driven wireless sensor networks.

Research on the Rapid Recognition Method of Electric Bicycles in Elevators Based on Machine Vision

People are gradually coming around to the idea of living a low-carbon lifestyle and using green transportation, and given the severe urban traffic congestion, electric bicycle commuting has taken over as the preferred mode of short-distance transportation for many. Since batteries are used to power electric bicycles, there are no greenhouse gas emissions while they are in use, which is more in line with the requirement for sustainable development around the world. The public has been increasingly concerned about the safety issues brought on by electric bicycles as a result of the industry’s quick development and the rapid increase in the number of electric bicycles worldwide. The unsafe operation of the elevator and the safety of the building have been seriously compromised by the unauthorized admission of electric bicycles into the elevator. To meet the need for fast detection and identification of electric bicycles in elevators, we designed a modified YOLOv5-based identification approach in this study. We propose the use of the EIoU loss function to address the occlusion problem in electric bicycle recognition. By considering the interaction ratio and overlap loss of the target frames, we are able to enhance localization accuracy and reduce the missed detection rate of occluded targets. Additionally, we introduce the CBAM attention mechanism in both the backbone and head of YOLOv5 to improve the expressive power of feature maps. This allows the model to prioritize important regions of the target object, leading to improved detection accuracy. Furthermore, we utilize the CARAFE operator during upsampling instead of the nearest operator in the original model. This enables our model to recover details and side information more accurately, resulting in finer sampling results. The experimental results demonstrate that our improved model achieves an mAP of 86.35 percent, a recall of 81.8 percent, and an accuracy of 88.0 percent. When compared to the original model under the same conditions, our improved YOLOv5 model shows an average detection accuracy increase of 3.49 percent, a recall increase of 5.6 percent, and an accuracy increase of 3.5 percent. Tests in application scenarios demonstrate that after putting the model on the hardware platform Jeston TX2 NX, stable and effective identification of electric bicycles can be accomplished.