Real-time Detection Algorithm Research Articles

Real-time jellyfish classification and detection algorithm based on improved YOLOv4-tiny and improved underwater image enhancement algorithm

The outbreak of jellyfish blooms poses a serious threat to human life and marine ecology. Therefore, jellyfish detection techniques have earned great interest. This paper investigates the jellyfish detection and classification algorithm based on optical images and deep learning theory. Firstly, we create a dataset comprising 11,926 images. A MSRCR underwater image enhancement algorithm with fusion is proposed. Finally, an improved YOLOv4-tiny algorithm is proposed by incorporating a CBMA module and optimizing the training method. The results demonstrate that the detection accuracy of the improved algorithm can reach 95.01%, the detection speed is 223FPS, both of which are better than the compared algorithms such as YOLOV4. In summary, our method can accurately and quickly detect jellyfish. The research in this paper lays the foundation for the development of an underwater jellyfish real-time monitoring system.

Retracted: A Real-Time Detection Algorithm of Flame Target Image.

[This retracts the article DOI: 10.1155/2022/5277805.].

MSDD-YOLOX: An enhanced YOLOX for real-time surface defect detection of oranges by type

Using an online high-throughput detection system for sorting oranges during the post-harvest process helped improve the commercialization level of the oranges industry. Surface defects on oranges created a poor first impression for consumers, making the rapid detection of oranges surface defects a primary concern for online sorting systems. However, due to variations in defect size and the visual similarity of different defects, there were still some challenges in detecting and identifying various surface defects on orange fruits based on their types. To address these challenges, this study first categorized surface defects on oranges into three major categories: deformity, scarring, and disease spot, based on their causes and potential post-harvest losses. Subsequently, to achieve real-time detection of orange surface defects on the orange sorting machine, a YOLOX-based real-time multi-type surface defect detection algorithm (MSDD-YOLOX) for oranges was proposed. This algorithm significantly improved the detection effectiveness of scarring at different scales by introducing neck network residual connections and cascading of the neck network. To address the issue of missed detections in texture-based defects and improve the regression of predicted bounding boxes, focal loss and Complete-IoU (CIoU) were employed in the algorithm. The results showed that MSDD-YOLOX achieved F1 values of 88.3 %, 80.4 %, and 92.7 % for the detection of deformity, scarring, and disease spot, respectively, with an overall detection F1 value of 90.8 %. These values represented improvements of 13.1 %, 10.2 %, 4.5 %, and 6.4 %, respectively, compared to the baseline model. Furthermore, compared to other deep learning object detectors, namely Faster RCNN, RetinaNet, FCOS, and Swin-Transformer, the proposed algorithm achieved optimal detection accuracy. Additionally, the MSDD-YOLOX model had a compact size of only 8.98 M, enabling real-time detection on the fruit grading line with an inference speed of up to 64.2FPS. Another innovation of this research was the external validation conducted on green oranges from Hainan and mandarins from Zhejiang. The results of external testing demonstrated that MSDD-YOLOX achieved overall F1 values of 90.6 % and 81.1 % for citrus fruits in these two regions, effectively proving the online deployment capability of MSDD-YOLOX and providing a robust solution for external defect detection in citrus fruits.

Research on Real-Time Detection Algorithm for Pavement Cracks Based on SparseInst-CDSM

This paper proposes a road crack detection algorithm based on an improved SparseInst network, called the SparseInst-CDSM algorithm, aimed at solving the problems of low recognition accuracy and poor real-time detection of existing algorithms. The algorithm introduces the CBAM module, DCNv2 convolution, SPM strip pooling module, MPM mixed pooling module, etc., effectively improving the integrity and accuracy of crack recognition. At the same time, the central axis skeleton of the crack is extracted using the central axis method, and the length and maximum width of the crack are calculated. In the experimental comparison under the self-built crack dataset, SparseInst-CDSM has an accuracy of 93.66%, a precision of 67.35%, a recall of 66.72%, and an IoU of 84.74%, all higher than mainstream segmentation models such as Mask-RCNN and SOLO that were compared, reflecting the superiority of the algorithm proposed in this paper. The comparison results of actual measurements show that the algorithm error is within 10%, indicating that it has high effectiveness and practicality.

An improved real-time detection algorithm based on frequency interpolation

Real-time monitoring of heart rate (HR), i.e., extraction of heart rate variability (HRV), plays an important role in diagnosis and prevention of cardiovascular diseases. Compared with traditional contact monitoring devices, the use of continuous wave (CW) Doppler radar to monitor HRV does not require contact and is not sensitive to light and temperature, which makes it more and more popular. To monitor the HRV based on CW Doppler radar, the time window must be shortened to less than 5 s, which will lead to the spectrum leakage and degrade the measurement accuracy of HRV. To solve this problem, a custom CW Doppler radar has been developed in an integrated fashion on a single PCB, whose transmitting frequency and power of the radar are 24 GHz and 3 dBm, respectively. Furthermore, four frequency interpolation algorithms are introduced to compare their extraction accuracy. Experiments are performed on three subjects, and results show that the Quinn algorithm can obtain best HRV extraction results compared with other algorithms. Specially, the average HRV extraction error is 3.61% using the Quinn algorithm.

A Real-Time Lightweight Detection Algorithm for Deck Crew and the Use of Fishing Nets Based on Improved YOLOv5s Network

A Real-Time Lightweight Detection Algorithm for Deck Crew and the Use of Fishing Nets Based on Improved YOLOv5s Network

YOLOv5-ASFF: A Multistage Strawberry Detection Algorithm Based on Improved YOLOv5

The smart farm is currently a hot topic in the agricultural industry. Due to the complex field environment, the intelligent monitoring model applicable to this environment requires high hardware performance, and there are difficulties in realizing real-time detection of ripe strawberries on a small automatic picking robot, etc. This research proposes a real-time multistage strawberry detection algorithm YOLOv5-ASFF based on improved YOLOv5. Through the introduction of the ASFF (adaptive spatial feature fusion) module into YOLOv5, the network can adaptively learn the fused spatial weights of strawberry feature maps at each scale as a way to fully obtain the image feature information of strawberries. To verify the superiority and availability of YOLOv5-ASFF, a strawberry dataset containing a variety of complex scenarios, including leaf shading, overlapping fruit, and dense fruit, was constructed in this experiment. The method achieved 91.86% and 88.03% for mAP and F1, respectively, and 98.77% for AP of mature-stage strawberries, showing strong robustness and generalization ability, better than SSD, YOLOv3, YOLOv4, and YOLOv5s. The YOLOv5-ASFF algorithm can overcome the influence of complex field environments and improve the detection of strawberries under dense distribution and shading conditions, and the method can provide technical support for monitoring yield estimation and harvest planning in intelligent strawberry field management.

Real-time dense small object detection algorithm based on multi-modal tea shoots.

The difficulties in tea shoot recognition are that the recognition is affected by lighting conditions, it is challenging to segment images with similar backgrounds to the shoot color, and the occlusion and overlap between leaves. To solve the problem of low accuracy of dense small object detection of tea shoots, this paper proposes a real-time dense small object detection algorithm based on multimodal optimization. First, RGB, depth, and infrared images are collected form a multimodal image set, and a complete shoot object labeling is performed. Then, the YOLOv5 model is improved and applied to dense and tiny tea shoot detection. Secondly, based on the improved YOLOv5 model, this paper designs two data layer-based multimodal image fusion methods and a feature layerbased multimodal image fusion method; meanwhile, a cross-modal fusion module (FFA) based on frequency domain and attention mechanisms is designed for the feature layer fusion method to adaptively align and focus critical regions in intra- and inter-modal channel and frequency domain dimensions. Finally, an objective-based scale matching method is developed to further improve the detection performance of small dense objects in natural environments with the assistance of transfer learning techniques. The experimental results indicate that the improved YOLOv5 model increases the mAP50 value by 1.7% compared to the benchmark model with fewer parameters and less computational effort. Compared with the single modality, the multimodal image fusion method increases the mAP50 value in all cases, with the method introducing the FFA module obtaining the highest mAP50 value of 0.827. After the pre-training strategy is used after scale matching, the mAP values can be improved by 1% and 1.4% on the two datasets. The research idea of multimodal optimization in this paper can provide a basis and technical support for dense small object detection.

A novel wearable device for automated real-time detection of epileptic seizures

BackgroundEpilepsy is a neurological disorder that has a variety of origins. It is caused by hyperexcitability and an imbalance between excitation and inhibition, which results in seizures. The World Health Organization (WHO) and its partners have classified epilepsy as a major public health concern. Over 50 million individuals globally are affected by epilepsy which shows that the patient’s family, social, educational, and vocational activities are severely limited if seizures are not controlled. Patients who suffer from epileptic seizures have emotional, behavioral, and neurological issues. Alerting systems using a wearable sensor are commonly used to detect epileptic seizures. However, most of the devices have no multimodal systems that increase sensitivity and lower the false discovery rate for screening and intervention of epileptic seizures. Therefore, the objective of this project was, to design and develop an efficient, economical, and automatically detecting epileptic seizure device in real-time.MethodsOur design incorporates different sensors to assess the patient’s condition such as an accelerometer, pulsoxymeter and vibration sensor which process body movement, heart rate variability, oxygen denaturation, and jerky movement respectively. The algorithm for real-time detection of epileptic seizures is based on the following: acceleration increases to a higher value of 23.4 m/s2 or decreases to a lower value of 10 m/s2 as energy is absorbed by the body, the heart rate increases by 10 bpm from the normal heart rate, oxygen denaturation is below 90% and vibration should be out of the range of 3 Hz -17 Hz. Then, a pulsoxymeter device was used as a gold standard to compare the heart rate variability and oxygen saturation sensor readings. The accuracy of the accelerometer and vibration sensor was also tested by a fast-moving and vibrating normal person’s hand.ResultsThe prototype was built and subjected to different tests and iterations. The proposed device was tested for accuracy, cost-effectiveness and ease of use. An acceptable accuracy was achieved for the accelerometer, pulsoxymeter, and vibration sensor measurements, and the prototype was built only with a component cost of less than 40 USD excluding design, manufacturing, and other costs. The design is tested to see if it fits the design criteria; the results of the tests reveal that a large portion of the scientific procedures utilized in this study to identify epileptic seizures is effective.ConclusionThis project is objectively targeted to design a medical device with multimodal systems that enable us to accurately detect epileptic seizures by detecting symptoms commonly associated with an episode of epileptic seizure and notifying a caregiver for immediate assistance. The proposed device has a great impact on reducing epileptic seizer mortality, especially in low-resource settings where both expertise and treatment are scarce.

A wavelet-based real-time fire detection algorithm with multi-modeling framework

This paper presents a wavelet-based real-time automated fire detection algorithm that takes into consideration the multi-resolution property of the wavelet transforms. Unlike conventional fire detection algorithms, which fail to capture temporal dependency within the fire sensor signals, the proposed wavelet-based features characterize temporal dynamics of chemical sensor signals generated from various types of fire, such as flaming, heating and smoldering fires. We propose a new feature selection technique based on types of fire to select the best features that can effectively discriminate between normal and various fire conditions. Then, a real-time fire detection algorithm with a multi-modeling framework is developed to effectively utilize the selected features and construct multiple fire detectors that are sensitive in monitoring various kinds of fires without prior knowledge. In addition, we develop a novel multi-sensor fusion system that incorporates various chemical sensors and collects an accurate and reliable fire dataset from different real-life fire scenarios in order to validate the performance of the proposed and existing fire detection algorithms. The experimental results with real-life and public fire data show that the proposed algorithm outperforms others with early detection time with a reasonable false alarm rate regardless of the type of fire.

An automated optical inspection (AOI) platform for three-dimensional (3D) defects detection on glass micro-optical components (GMOC)

With the widespread deployment of wavelength division multiplexing (WDM), optical transceivers increasingly use many glass micro-optical components (GMOC). Visual inspection of these GMOCs is a critical manufacturing step to ensure quality and reliability. However, manual inspection is often labor-intensive and time-consuming due to the transparent nature of glass components and the small, randomly located defects in three dimensions. Although automated optical inspection (AOI) exists, it has not yet been able to provide the desired level of accuracy and efficiency.This paper reports the development of an AOI platform for 3D defect detection on GMOCs. The platform incorporates 3D video acquisition and a novel two-stage neural network machine-learning algorithm. It includes a robotic arm for moving parts in 3D, a camera with an illumination module for video acquisition, and a video streaming processing unit with a machine vision algorithm for real-time defect detection on a production line. The robotic arm enables multi-perspective video capture of a test sample without refocusing. The two-stage machine learning network uses a modified YOLOv4 architecture with color channel separation (CCS) convolution, an image quality evaluation (IQE) module, and a frame fusion module to integrate the single frame detection results. This network can process multi-perspective video streams in real-time for defects detection in a coarse-to-fine manner. The AOI platform was trained with only 30 samples and achieved promising performances with a recall rate of 1, a detection accuracy of 97%, and an inspection time of 48 s per part.

Mobile Robot System for Selective Asparagus Harvesting

Asparagus harvesting presents unique challenges, due to the variability in spear growth, which makes large-scale automated harvesting difficult. This paper describes the development of an asparagus harvesting robot system. The system consists of a delta robot mounted on a mobile track-based platform. It employs a real-time asparagus detection algorithm and a sensory system to determine optimal harvesting points. Low-level control and high-level control are separated in the robot control. The performance of the system was evaluated in a laboratory field mock-up and in the open field, using asparagus spears of various shapes. The results demonstrate that the system detected and harvested 88% of the ready-to-harvest spears, with an average harvesting cycle cost of 3.44s±0.14s. In addition, outdoor testing in an open field demonstrated a 77% success rate in identifying and harvesting asparagus spears.

Facial Emotion Recognition System for Isolated Patient Monitoring using ResNet-50 based Convolutional Neural Network Model

Facial emotion recognition is a promising technology that has the potential to improve the quality of care for isolated patients by enabling healthcare professionals to assess the emotional state of these patients more accurately. Isolated patient monitoring is the practice of closely observing patients who have been isolated due to the risk of infection or other health concerns, and it is essential to ensure the well-being of these patients. The Facial Expression Recognition system is the process of identifying the emotional state of a person. In this system, captured image is compared with the trained dataset available in database and then emotional state of the image will be displayed. This will also integrate the system into existing patient monitoring systems and collect data on its effectiveness in improving the care and well-being of isolated patients. The ultimate goal is to demonstrate the value of facial emotion recognition in isolated patient monitoring, and to identify any potential ethical or privacy concerns that may need to be addressed. This is for facial emotion recognition (FER) in isolated patient monitoring. FER is achieved using computer vision techniques and voice recognition techniques to analyze and interpret facial expressions, which can be useful for identifying and tracking changes in patient’s emotional states. The proposed FER system utilizes deep learning algorithms for real-time detection and classification of emotions with high accuracy. Two popular methods utilized mostly in the literature for the automatic FER systems are based on geometry and appearance. Facial Expression Recognition usually performed in three-stages consisting of pre-processing, face detection & feature extraction, and expression classification. The paper analyses various deep learning image classification models (convolutional neural networks) to identify the certain human emotions: depressed, despair, fear, irritated, uneasy, shocked and neutrality.

Применение нейронных сетей для распознавания объектов на железнодорожном транспорте

Purpose: With the help of vision systems and neural networks, such as YOLOv8 and MASK R-CNN, it is possible to quickly and accurately detect objects that can lead to an accident or delay trains. YOLOv8 is one of the most popular real-time object detection algorithms that uses deep neural networks to classify and localize objects. YOLOv8 can detect objects in images and videos with high speed and accuracy. This model can work on various hardware platforms, including mobile devices and computers. MASK R-CNN is an even more advanced object detection algorithm that has the ability to highlight objects and their contours with high accuracy. MASK R-CNN uses convolutional neural networks and mask segmentation techniques to detect objects. It can work both in real time and on static images. When vision systems are equipped with YOLOv8 and MASK R-CNN neural networks, they can quickly respond to extraneous objects that appear on the rails. The purpose of the article is to develop algorithms for detecting railway transport objects and obstacles using technical vision and neural networks, and to evaluate the effectiveness of algorithms. Methods: The YOLOv8 algorithm is based on the architecture of convolutional neural networks and uses supervised learning methods. This model takes an image as input and provides estimates of the probability that a certain object is present in the image in real time. To achieve this, YOLOv8 employs region of interest (ROI) detection methods, allowing to determine the areas of the image on which objects may be located. The MASK R-CNN algorithm uses more sophisticated methods, such as mask segmentation methods and proportional resizing of the area of interest (RoIAlign) to achieve more accurate results of object detecting in images and videos. It is also based on convolutional neural networks and uses supervised learning methods. MASK R-CNN uses mask segmentation methods to determine the contour of an object in an image, as well as the RoIAlign method, which allows for superior quality when processing various image sizes. Common mathematical methods that are used in YOLOv8 and MASK R-CNN are methods of convolutional neural network, supervised learning and optimization of the loss function. They are based on deep learning algorithms such as stochastic gradient descent and backward propagation of errors. Results: An algorithm for detecting foreign objects on the route of rolling stock using a technical vision system, calculation of the evaluation of the quality of neural networks performance, error matrices have been formed, the results of neural network processing have been obtained. Practical significance: An algorithm for detecting foreign objects on the route of the moving rolling stock using a technical vision system has been developed, two neural networks have been trained to detect railway transport objects and obstacles on the way.

Lightweight real-time lane detection algorithm based on ghost convolution and self batch normalization

Lightweight real-time lane detection algorithm based on ghost convolution and self batch normalization

Realization of Autonomous Detection, Positioning and Angle Estimation of Harvested Logs

To further develop forest production, higher automation of forest operations is required. Such endeavour promotes research on unmanned forest machines. Designing unmanned forest machines that exercise forwarding requires an understanding of positioning and angle estimations of logs after cutting and delimbing have been conducted, as support for subsequent crane loading work. This study aims to improve the automation of the forwarding operation and presents a system to realize real-time automatic detection, positioning, and angle estimation of harvested logs implemented on an existing unmanned forest machine experimental platform from the AORO (Arctic Off-Road Robotics) Lab. This system uses ROS as the underlying software architecture and a Zed2 camera and NVIDIA JETSON AGX XAVIER as the imaging sensor and computing platform, respectively, utilizing the YOLOv3 algorithm for real-time object detection. Moreover, the study combines the processing of depth data and depth to spatial transform to realize the calculation of the relative location of the target log related to the camera. On this basis, the angle estimation of the target log is further realized by image processing and color analysis. Finally, the absolute position and log angles are determined by the spatial coordinate transformation of the relative position data. This system was tested and validated using a pre-trained log detector for birch with a mean average precision (mAP) of 80.51%. Log positioning mean error did not exceed 0.27 m and the angle estimation mean error was less than 3 degrees during the tests. This log pose estimation method could encompass one important part of automated forwarding operations.

Image Detection and Real Time Object Detection

Abstract: Object detection is a crucial task in computer vision with various practical applications, including surveillance, autonomous vehicles, and robotics. The YOLO (You Only Look Once) algorithm is a popular real-time object detection algorithm that has gained significant attention due to its high accuracy and speed. This algorithm processes the entire image at once and predicts bounding boxes and class probabilities for identified objects, making it ideal for time-sensitive applications. YOLO has evolved through various versions, with YOLOv5 being the latest and most advanced version that employs a feature pyramid network (FPN) and anchor boxes to improve its object detection accuracy. In this project, we aim to implement YOLOv5 for real-time object detection and image detection tasks. We will train the model on a suitable dataset and evaluate its performance on various benchmarks, comparing it with other advanced object detection algorithms. The project's outcome will provide a robust and efficient solution for real-time object detection that can aid quick decision-making in identifying object categories and their respective positions. It has practical applications in surveillance, automated driving, and robotics.

Vision-based underwater target real-time detection for autonomous underwater vehicle subsea exploration

Autonomous underwater vehicles (AUVs) equipped with online visual inspection systems can detect underwater targets during underwater operations, which is of great significance to subsea exploration. However, the undersea scene has some instinctive challenging problems, such as poor lighting conditions, sediment burial, and marine biofouling mimicry, which makes it difficult for traditional target detection algorithms to achieve online, reliable, and accurate detection of underwater targets. To solve the above issues, this paper proposes a real-time object detection algorithm for underwater targets based on a lightweight convolutional neural network model. To improve the imaging quality of underwater images, contrast limited adaptive histogram equalization with the fused multicolor space (FCLAHE) model is designed to enhance the image quality of underwater targets. Afterwards, a spindle-shaped backbone network is designed. The inverted residual block and group convolutions are used to extract depth features to ensure the target detection accuracy on one hand and to reduce the model parameter volume on the other hand under complex scenarios. Through extensive experiments, the precision, recall, and mAP of the proposed algorithm reached 91.2%, 90.1%, and 88.3%, respectively. It is also noticeable that the proposed method has been integrated into the embedded GPU platform and deployed in the AUV system in the practical scenarios. The average computational time is 0.053s, which satisfies the requirements of real-time object detection.

An Improved Lightweight Real-Time Detection Algorithm Based on the Edge Computing Platform for UAV Images

Unmanned aerial vehicle (UAV) image detection algorithms are critical in performing military countermeasures and disaster search and rescue. The state-of-the-art object detection algorithm known as you only look once (YOLO) is widely used for detecting UAV images. However, it faces challenges such as high floating-point operations (FLOPs), redundant parameters, slow inference speed, and poor performance in detecting small objects. To address the above issues, an improved, lightweight, real-time detection algorithm was proposed based on the edge computing platform for UAV images. In the presented method, MobileNetV3 was used as the YOLOv5 backbone network to reduce the numbers of parameters and FLOPs. To enhance the feature extraction ability of MobileNetV3, the efficient channel attention (ECA) attention mechanism was introduced into MobileNetV3. Furthermore, in order to improve the detection ability for small objects, an extra prediction head was introduced into the neck structure, and two kinds of neck structures with different parameter scales were designed to meet the requirements of different embedded devices. Finally, the FocalEIoU loss function was introduced into YOLOv5 to accelerate bounding box regression and improve the localization accuracy of the algorithm. To validate the performance of the proposed improved algorithm, we compared our algorithm with other algorithms in the VisDrone-Det2021 dataset. The results showed that compared with YOLOv5s, MELF-YOLOv5-S achieved a 51.4% reduction in the number of parameters and a 38.6% decrease in the number of FLOPs. MELF-YOLOv5-L had 87.4% and 47.4% fewer parameters and FLOPs, respectively, and achieved higher detection accuracy than YOLOv5l.

Real-time fire detection algorithms running on small embedded devices based on MobileNetV3 and YOLOv4

AimFires are a serious threat to people’s lives and property. Detecting fires quickly and effectively and extinguishing them in the nascent stage is an effective way to reduce fire hazards. Currently, deep learning-based fire detection algorithms are usually deployed on the PC side.MethodsAfter migrating to small embedded devices, the accuracy and speed of recognition are degraded due to the lack of computing power. In this paper, we propose a real-time fire detection algorithm based on MobileNetV3-large and yolov4, replacing CSP Darknet53 in yolov4 with MobileNetV3-large to achieve the initial extraction of flame and smoke features while greatly reducing the computational effort of the network structure. A path connecting PANet was explored on Gbneck(104, 104, 24), while SPP was embedded in the path from MobileNetV3 to PANet to improve the feature extraction capability for small targets; the PANet in yolo4 was improved by combining the BiFPN path fusion method, and the improved PANet further improved the feature extraction capability; the Vision Transformer model is added to the backbone feature extraction network and PANet of the YOLOv4 model to give full play to the model’s multi-headed attention mechanism for pre-processing image features; adding ECA Net to the head network of yolo4 improves the overall recognition performance of the network.ResultThese algorithms run well on PC and reach 95.14% recognition accuracy on the public dataset BoWFire. Finally, these algorithms were migrated to the Jeston Xavier NX platform, and the entire network was quantized and accelerated with the TensorRT algorithm. With the image propagation function of the fire robot, the overall recognition frame rate can reach about 26.13 with high real-time performance while maintaining a high recognition accuracy.ConclusionSeveral comparative experiments have also validated the effectiveness of this paper’s improvements to the YOLOv4 algorithm and the superiority of these structures. With the effective integration of these components, the algorithm shows high accuracy and real-time performance.