Spot Detection Research Articles

Real-time laser spot detection and tracking system based on parallel multi-target detection and determination algorithm.

Laser spot detection and tracking play a critical role in laser techniques. However, traditional detection and tracking systems tend to be bulky and lack portability. Therefore, there is a growing emphasis on developing high-performance and miniaturized systems based on the field programmable gate array (FPGA). In this paper, a novel parallel multi-target detection and determination algorithm is proposed to address the issue of current FPGA-based systems' ineffective detection of laser spots in complex environments. Our simulation results demonstrate that the algorithm can effectively detect laser spots in complex environments. It can process a frame with an 800 × 480 resolution in only 7.88ms at a 50MHz image processing frequency, which means it can process more than 100 f/s and meet the real-time detection requirements. Such excellent detection performance is challenging to achieve with central processing units and advanced RISC machine microprocessors. Then, the algorithm is further deployed on an FPGA to build a prototype laser spot detection and tracking system. Practical tests show that the system can achieve a spot detection accuracy of around 90% under different luminous intensities, indicating excellent robustness of the designed algorithm. Besides, with the use of a piezoelectric actuator, speedy and precise tracking of the laser spot is implemented. The characteristics of speedy response, self-latching in power off, and no electromagnetic interference of the piezoelectric actuator give the system tremendous advantages in developing high-precision wireless communication control technology, which further broadens the application of the proposed system.

Identification of walnut (Juglans regia) leaf spot agent by its production of species-specific phytotoxin

Aim. For the purpose of correct species and internal identification of isolated strains of bacteria that cause walnut leaf spotting, molecular biological and genetic detection of their ability to synthesize the phytotoxin syringomycin have been carried out. Methods. It has been used microbiological and molecular genetic (PCR) methods. Results. The presence of syringomycin (syrD) phytotoxin secretion genes was established in isolated strains of walnut leaf spot pathogen and the typical strain of Pseudomonas syringae pv. syringae UCM B-1027Т. Biotesting of the ability to synthesize syringomycin by these strains revealed the similarity of the antagonistic action of the strains isolated from walnut and the typical strain of Pseudomonas syringae pv. syringae UCM B-1027T, which indicates their ability to synthesize lipodepsinopeptide phytotoxin – syringomycin. Conclusions. Based on the results of the previous researches and presented in this work, strains isolated from walnut were identified as a members of species Pseudomonas syringae pv. syringae.

Effects of Advanced Driver Assistance Systems on Impact Velocity and Injury Severity: An Exploration of Data from the Crash Investigation Sampling System

This paper presents an exploratory analysis of advanced driver assistance systems (ADAS) features and safety outcomes using National Highway Transportation Safety Administration’s Crash Investigation Sampling System (CISS) database. In this database, we found improved safety outcomes associated with the presence of three ADAS features: lane departure warning, forward collision warning, and blind spot detection. In particular, vehicles with ADAS features were less likely to be involved in fatal and severe crashes than vehicles of a similar age and body style. Additionally, vehicles with ADAS features had a lower rate of change in velocity pre- and postcrash, a key proxy for injury severity. Vehicles with ADAS were also less likely to be involved in head-on and rear-end crashes than similar vehicles without ADAS and were less likely to experience a severe outcome than vehicles without ADAS when involved in sideswipe collisions. Despite these promising findings, however, small sample sizes precluded robust within-category and multivariate analysis. More robust data are urgently needed to disentangle the safety effects of ADAS features from those related to confounding variables such as vehicle age and body type for vehicle occupants, pedestrians, and bicyclists. We conclude the paper with recommendations for improvements to the CISS database that could support future investigations. We hope this work provides a baseline for scholars and practitioners to deepen our understanding of the effects of ADAS features on transportation safety.

An Effective YOLO-Based Proactive Blind Spot Warning System for Motorcycles

Interest in advanced driver assistance systems (ADAS) is booming in recent years. One of the most effervescent ADAS features is blind spot detection (BSD), which uses radar sensors or cameras to detect vehicles in the blind spot area and alerts the driver to avoid a collision when changing lanes. However, this kind of BSD system fails to notify nearby vehicle drivers in this blind spot of the possible collision. The goal of this research is to design a proactive bus blind spot warning (PBSW) system that will immediately notify motorcyclists when they enter the blind spot or the area of the inner wheel difference of a target vehicle, i.e., a bus. This will increase the real-time functionality of BSD and can have a significant impact on enhancing motorcyclist safety. The proposed hardware is placed on the motorcycle and consists of a Raspberry Pi 3B+ and a dual-lens stereo camera. We use dual-lens cameras to capture and create stereoscopic images then transmit the images from the Raspberry Pi 3B+ to an Android phone via Wi-Fi and to a cloud server using a cellular network. At the cloud server, we use the YOLOv4 image recognition model to identify the position of the rear-view mirror of the bus and use the lens imaging principle to estimate the distance between the bus and the motorcyclist. Finally, the cloud server returns the estimated distance to the PBSW app on the Android phone. According to the received distance value, the app will display the visible area/blind spot, the area of the inner wheel difference of the bus, the position of the motorcyclist, and the estimated distance between the motorcycle and the bus. Hence, as soon as the motorcyclist enters the blind spot of the bus or the area of the inner wheel difference, the app will alert the motorcyclist immediately to enhance their real-time safety. We have evaluated this PBSW system implemented in real life. The results show that the average position accuracy of the rear-view mirror is 92.82%, the error rate of the estimated distance between the rear-view mirror and the dual-lens camera is lower than 0.2%, and the average round trip delay between the Android phone and the cloud server is about 0.5 s. To the best of our knowledge, this proposed system is one of few PBSW systems which can be applied in the real world to protect motorcyclists from the danger of entering the blind spot and the area of the inner wheel difference of the target vehicle in real time.

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.

Juno's Multi‐Instruments Observations During the Flybys of Auroral Bright Spots in Jupiter's Polar Aurorae

AbstractJuno's arrival at Jupiter in 2016 revealed unprecedented details about Jupiter's ultraviolet aurorae thanks to its unique suite of remote sensing and in situ instruments. Here we present results from in situ observations during Juno flybys above specific bright auroral spots in Jupiter's polar aurora. We compare data observed by Juno‐UVS, JEDI, JADE, Waves, and MAG instruments when Juno was magnetically connected to bright polar auroral spots (or their immediate vicinity) during perijove 3 (PJ3), PJ15, and PJ33. The highly energetic particles observed by JEDI show enhancements dominated by upward electrons, which suggests that the particle acceleration region takes place below the spacecraft. Moreover, brightness and upward particle flux were higher for the northern bright spot in PJ3 than the southern spots found in PJ15 and PJ33. In addition, we notice the intensification of whistler‐mode waves at the time of the particle enhancements, suggesting that wave‐particle interactions contribute to the acceleration of particles that cause the UV aurorae. The MAG data reveal magnetic perturbations during the PJ3 spot detection by Juno, which suggests the presence of significant field‐aligned electric currents. While the stable positions of the bright spots in System III suggest that the phenomenon is fixed to the planet's rotation, the presence of field‐aligned currents leaves the possibility of an origin rooted much farther in the magnetosphere.

IoT System Based on Artificial Intelligence for Hot Spot Detection in Photovoltaic Modules for a Wide Range of Irradiances.

Infrared thermography (IRT) is a technique used to diagnose Photovoltaic (PV) installations to detect sub-optimal conditions. The increase of PV installations in smart cities has generated the search for technology that improves the use of IRT, which requires irradiance conditions to be greater than 700 W/m2, making it impossible to use at times when irradiance goes under that value. This project presents an IoT platform working on artificial intelligence (AI) which automatically detects hot spots in PV modules by analyzing the temperature differentials between modules exposed to irradiances greater than 300 W/m2. For this purpose, two AI (Deep learning and machine learning) were trained and tested in a real PV installation where hot spots were induced. The system was able to detect hot spots with a sensitivity of 0.995 and an accuracy of 0.923 under dirty, short-circuited, and partially shaded conditions. This project differs from others because it proposes an alternative to facilitate the implementation of diagnostics with IRT and evaluates the real temperatures of PV modules, which represents a potential economic saving for PV installation managers and inspectors.

Living-skin Detection using Multi-layer Skin Property Perceived by the Structured Light.

Living-skin detection has been used to prevent the attack of face fraud in a face recognition system. In this paper, we propose a new concept that exploits the multi-layer structure property of skin for living-skin detection. We observe a significant difference in the blur of the laser spot created by the structured light on the skin and non-skin due to the characteristic properties of laser photons in skin penetration and reflection. Based on this observation, we designed a new living-skin detection algorithm to differentiate skin and non-skin based on the blur detection of laser spots. The experimental results show that the proposed setup and method have a promising performance with an averaged precision of 96.7%, averaged recall of 82.2%, and averaged F1-score of 88.6% on a dataset of 20 adult subjects. This demonstrates the effectiveness of the new concept that uses multi-layer properties of skin tissues for living-skin detection, which may lead to new solutions for face anti-spoofing.

Reweighted L1 Minimization for Networkwide Weak Spot Detection from Traffic Speed Deflectometer Measurements

The traffic speed deflectometer (TSD) can collect network-level deflection data in a cost-effective and timely manner. However, an automated feature extraction method is needed to interpret such large amounts of data. Basis pursuit (BP) is one such technique: BP can sparsely decompose TSD measurements over a given basis or set of bases of the signal vector space that can represent a particular signal feature. For instance, the TSD surface deflection estimates can be reconstructed as a combination of wavelets, which represent continuously varying features like changes in pavement properties, plus pulses representing the response from structurally weak spots. Yet, the denoised measurements estimated by BP may either be riddled with several false positives (spikes that mismatch real weak spots) or be constructed out of true positive features with damped amplitude. This paper presents reweighed L1 minimization (RWL1), an enhancement to BP to both correct the dampening and discard false positives. This paper introduces RWL1 with examples from simulated data to show its advantage over vanilla BP denoising, plus a demonstration featuring real TSD measurements from a networkwide survey to demonstrate RWL1’s potential as an exploratory analysis tool to detect structurally weak locations within the pavement network worthy of further investigation at the project level.

Intelligent Application of Laser for Medical Prognosis: An Instance for Laser Mark Diabetic Retinopathy

ABSTRACT: Refractive laser surgery is all about the accuracy, whether screening or surgery, given the age and profile of the patient enduring these trials, there is no margin for error. Most of them are for aesthetic reasons, contact lens intolerance, or professional reasons, including athletes. In this article, the role of artificial intelligence and deep learning in laser eye surgeries has been introduced. The presence of lingering laser spots on the retina after refractive laser surgery in diabetic retinopathy poses a potential risk to visual integrity and ocular well-being. The hypothesis for the research paper is that the hybridized convolutional neural network models, including LeNet-1, AlexNet, VGG16, PolyNet, Inception V2, and Inception-ResNetV2, will yield varying levels of performance in classifying and segmenting laser spots in the retina after diabetic retinopathy surgery. The hypothesis predicts that Inception-ResNetV2 will demonstrate superior results compared to the other CNN versions. The research aims to provide a novel approach for laser therapies and treatments, facilitating the rapid classification, highlighting, and segmentation of laser marks on the retina for prompt medical precautions. The comparative analysis revealed that Inception-ResNetV2 exhibited exceptional performance in both training and validation, achieving the highest accuracy (96.54%) for classifying diabetic retinopathy images. Notably, VGG16 also demonstrated strong performance with a validation accuracy of 94%. Conversely, LeNet-1, AlexNet, PolyNet, and Inception V2 displayed comparatively lower accuracy rates, suggesting their architectures may be less optimized for this particular image classification task. This achievement holds immense promise for timely detection, precise localization, and optimal management of laser spots, fostering enhanced visual outcomes and elevating the standards of patient care in this context.

Performance of heat spots obtained from satellite datasets to represent burned areas in Andean ecosystems of Cusco, Peru

The combustion of biomass is a prevalent practice in the Andes. Often, these burns escape control and escalate into wildfires. However, the investigation of unreported fire incidents has not received the same level of attention as reported wildfires. Satellite provide an alternative information source for studying wildfires. Given the scarcity of wildfire response tools, it is imperative to develop strategies to prevent burns in regions where human activity typically triggers wildfires. This study aims to assess the efficacy of heat spot data obtained from satellite datasets in identifying fire activity in the Peruvian Andes. The study utilized MODIS (MCD14DL product) and VIIRS (SUOMI and JPSS-1 products) satellite datasets to characterize fire activity through heat spot detection. Additionally, the study employed the normalized burned area index (NBR), a valuable indicator for mapping burned areas. Our findings indicate that MODIS and VIIRS heat spots demonstrate a high level of reliability in detecting active fires (commission errors of ∼1%). However, the detection of burned areas not captured by MODIS or VIIRS heat spots was significantly high (omission errors of ∼90%). Nevertheless, this detection deficiency decreased for larger burn areas (errors of omission between 10 and 30% for burned areas between 50 and 100 ha, using JPSS-1 VIIRS). These results suggest that satellite heat spots are inadequate for identifying burn practices, which encompass small-scale and short-duration fire activities (lasting only hours). The outcomes of this study enhance our understanding of the suitability of heat spot detection for wildfire prevention in the Peruvian Andes.

Whole-mount smFISH allows combining RNA and protein quantification at cellular and subcellular resolution

Multicellular organisms result from complex developmental processes largely orchestrated through the quantitative spatiotemporal regulation of gene expression. Yet, obtaining absolute counts of messenger RNAs at a three-dimensional resolution remains challenging, especially in plants, owing to high levels of tissue autofluorescence that prevent the detection of diffraction-limited fluorescent spots. In situ hybridization methods based on amplification cycles have recently emerged, but they are laborious and often lead to quantification biases. In this article, we present a simple method based on single-molecule RNA fluorescence in situ hybridization to visualize and count the number of mRNA molecules in several intact plant tissues. In addition, with the use of fluorescent protein reporters, our method also enables simultaneous detection of mRNA and protein quantity, as well as subcellular distribution, in single cells. With this method, research in plants can now fully explore the benefits of the quantitative analysis of transcription and protein levels at cellular and subcellular resolution in plant tissues.

Spatial Differentiation and Driving Force Detection of Rural Settlements in the Yangtze River Delta Region

As an economically developed region, the Yangtze River Delta region has undergone earth-shaking changes in its rural settlements due to rapid urbanization. For the optimization and adjustment of rural settlements, it is crucial to disclose their distinguishing spatial features and impelling factors. Taking 307 county-level administrative regions in the Yangtze River Delta region as the research object, this study comprehensively uses the landscape index, nearest neighbor index, Moran index, and spatial hot spot detection system to analyze the spatial differentiation characteristics of rural residential location-scale morphology and reveals its driving factors using the optimal parameters-based geographical detector model. According to the findings, rural settlements in the Yangtze River Delta region exhibit an average nearest neighbor index of 0.7417, a Moran’s I of 1.2993 for the number of patches (NP), and a maximum patch density (PD) of 17.25 villages per square kilometer. It has significant characteristics of large-scale village cluster distribution, and the morphology of rural settlements in the southern and northern regions shows apparent differences. The natural environment and social economies, such as elevation, slope, precipitation, and population density, mainly drive the location-scale morphological spatial distribution of rural settlements. At the same time, the interaction between the natural environment, social economy, and location condition factors has a synergistic enhancement effect on the spatial distribution of location-scale morphology of rural settlements.

Spatial and temporal distribution of reported dengue cases and hot spot identification in Quezon City, Philippines, 2010–2017

BackgroundDengue remains a major public health problem in the Philippines, particularly in urban areas of the National Capital Region. Thematic mapping using geographic information systems complemented by spatial analysis such as cluster analysis and hot spot detection can provide useful information to guide preventive measures and control strategies against dengue. Hence, this study was aimed to describe the spatiotemporal distribution of dengue incidence and identify dengue hot spots by barangay using reported cases from Quezon City, the Philippines from 2010 to 2017.MethodsReported dengue case data at barangay level from January 1, 2010 to December 31, 2017 were obtained from the Quezon City Epidemiology and Surveillance Unit. The annual incidence rate of dengue from 2010 to 2017, expressed as the total number of dengue cases per 10,000 population in each year, was calculated for each barangay. Thematic mapping, global cluster analysis, and hot spot analysis were performed using ArcGIS 10.3.1.ResultsThe number of reported dengue cases and their spatial distribution varied highly between years. Local clusters were evident during the study period. Eighteen barangays were identified as hot spots.ConclusionsConsidering the spatial heterogeneity and instability of hot spots in Quezon City across years, efforts towards the containment of dengue can be made more targeted, and efficient with the application of hot spot analysis in routine surveillance. This may be useful not only for the control of dengue but also for other diseases, and for public health planning, monitoring, and evaluation.

Blind Spot Detection Radar System Design for Safe Driving of Smart Vehicles

Recently, there has been extensive research and development in the field of smart cars, including technologies related to autonomous driving. Various industries are actively working towards creating efficient and safe self-driving cars. Sensor technologies are emerging to prevent traffic accidents and support safe driving in complex environments where human perception may be limited. One of the representative technologies being researched is the use of Frequency Modulated Continuous Wave (FMCW) radar. Automobile manufacturers are improving driving safety by equipping cars with Blind Spot Detection (BSD) radar systems that use FMCW technology. As the complexity of driving environments continues to grow, ongoing research is aimed at enhancing the accuracy and reliability of BSD radar technology for detecting blind spots in vehicles. This paper presents the signal processing and tracking algorithms that are the core technologies of the BSD radar, and the design for a BSD radar system. The designed radar system was installed on a vehicle to verify its performance in real-world road environments. The ultimate objective of this research is to design a BSD radar system with high accuracy and reliability in BSD detection using AI technology. In pursuit of this goal, this paper presents the hardware design of the BSD radar system, including antenna and modem designs.

A Method to Directly Identify Cronobacter sakazakii in Liquid Medium by MALDI-TOF MS.

Matrix-assisted laser desorption ionization time-of-flight mass spectrometry has been widely used as an emerging technology for the rapid identification of microorganisms. Cronobacter sakazakii (C. sakazakii) is a food-borne pathogen of particular importance to the powdered infant formula (PIF) processing environment due to its high lethality in infants. However, the traditional solid spotting detection method of pretreating samples for MALDI-TOF MS leads only to qualitative detection of C. sakazakii. We developed a new, low-cost, robust liquid spotting pretreatment method and used a response surface methodology to optimize its parameters. The applicability, accuracy, and quantitative potential were measured for different types of samples. The optimal parameters of this method were as follows: a volume of 70% formic acid of 25 μL, treatment with ultrasound at 350 W for 3 min, and a volume of acetonitrile added of 75 μL. These conditions led to the highest identification score for C. sakazakii (1926.42 ± 48.497). This method was found to detect bacteria accurately and reproducibly. When 70 strains of C. sakazakii isolates were analyzed with this method, the identification accuracy was 100%. The detection limit of C. sakazakii in environmental and PIF samples was 4.1 × 101 cfu/mL and 2.72 × 103 cfu/mL, respectively.

Structured Illumination Microscopy Improves Spot Detection Performance in Spatial Transcriptomics.

Spatial biology is a rapidly growing research field that focuses on the transcriptomic or proteomic profiling of single cells within tissues with preserved spatial information. Imaging-based spatial transcriptomics uses epifluorescence microscopy, which has shown remarkable results for the identification of multiple targets in situ. Nonetheless, the number of genes that can be reliably visualized is limited by the diffraction of light. Here, we investigate the effect of structured illumination (SIM), a super-resolution microscopy approach, on the performance of single-gene transcript detection in spatial transcriptomics experiments. We performed direct mRNA-targeted hybridization in situ sequencing for multiple genes in mouse coronal brain tissue sections. We evaluated spot detection performance in widefield and confocal images versus those with SIM in combination with 20×, 25× and 60× objectives. In general, SIM increases the detection efficiency of gene transcript spots compared to widefield and confocal modes. For each case, the specific fold increase in localizations is dependent on gene transcript density and the numerical aperture of the objective used, which has been shown to play an important role, especially for densely clustered spots. Taken together, our results suggest that SIM has the capacity to improve spot detection and overall data quality in spatial transcriptomics.

The SNR of Positron Emission Data With Gaussian and Non-Gaussian Time-of-Flight Kernels, With Application to Prompt Photon Coincidence.

It is well known that measurement of the time-of-flight (TOF) increases the information provided by coincident events in positron emission tomography (PET). This information increase propagates through the reconstruction and improves the signal-to-noise ratio in the reconstructed images. Takehiro Tomitani has analytically computed the gain in variance in the reconstructed image, provided by a particular TOF resolution, for the center of a uniform disk and for a Gaussian TOF kernel. In this paper we extend this result, by computing the signal-to-noise ratio (SNR) contributed by individual coincidence events for two different tasks. One task is the detection of a hot spot in the center of a uniform cylinder. The second one is the same as that considered by Tomitani, i.e. the reconstruction of the central voxel in the image of a uniform cylinder. In addition, we extend the computation to non-Gaussian TOF kernels. It is found that a modification of the TOF-kernel changes the SNR for both tasks in almost exactly the same way. The proposed method can be used to compare TOF-systems with different and possibly event-dependent TOF-kernels, as encountered when prompt photons, such as Cherenkov photons are present, or when the detector is composed of different scintillators. The method is validated with simple 2D simulations and illustrated by applying it to PET detectors producing optical photons with event-dependent timing characteristics.

A Small Sample Photovoltaic Hot Spot Identification Method Based on Deep Transfer Learning

A small-sample photovoltaic hot spot identification method based on deep transfer learning has been proposed as a solution to the problem that traditional deep learning models require a substantial amount of training data, whereas the number of hot spot effect samples is relatively low and difficult to collect. This issue can be resolved by the development of a solution that employs deep transfer learning. The Inception-v 4 model serves as the basis for building a deep transfer learning model. After finishing the training procedure using a small-sample hot spot dataset of negative sample multi-classification, a network model for hot spot detection is built. Experiments reveal that even when the number of accessible samples is restricted, the model trained using the deep transfer learning technique has great identification accuracy, a low false detection rate, and strong generalization capability.

Detection of brown spot and leaf blight in rice with the You Only Look Once (YOLO) algorithm

Rice is a plant that produces rice as the staple food for almost all Indonesians. Rice plays a vital role in the food chain in Indonesia. However, many types of diseases on rice leaves attack and inhibit the growth of rice in Indonesian rice fields. Among them are brown spots and leaf blight. So far, the classification of diseases on rice leaves is only based on the experience of farmers. This results in uncertainty and diagnostic inaccuracy in classifying various diseases on rice plants. In this case, a method is needed to solve the problem. The method used in this research is digital imagery by processing images of diseases on the leaves of rice plants using the YOLO (You Only Look Once) algorithm. YOLO uses a single Convolutional Neural Network (CNN) for object classification and localization using Bounding Box. This research aims to provide options to users in determining the diagnosis of leaf spots and blight diseases on the leaves of rice plants. The average Precision (mAP) evaluation result is 69%, indicating that this method is suitable for detecting diseases on rice leaves.