Level Classification Research Articles

Crowd Density Level Classification for Service Waiting Room Based on Head Detection to Enhance Visitor Experience

The crowd within a confined space can potentially lead to air stagnation in waiting areas. Constantly running air conditioning throughout the day to balance air circulation may result in excessive energy consumption by the building. To address this issue, Heating, Ventilating, and Air-Conditioning (HVAC) systems are employed to manage and regulate indoor energy usage. However, sensor-based detection often fails to capture human variables promptly, resulting in less accurate density readings. Camera footage proves to be more reliable than sensors in accurately detecting crowds. This research utilizes You Only Look Once version 8 (YOLOv8), a robust algorithm for object detection, particularly effective in crowd detection for images, along with Convolutional Vision Transformer (CvT) for crowd density level classification into "Normal" and "Crowded" levels. CvT enhances classification accuracy by incorporating function from Convolutional Neural Network (CNN) in model training, including receptive field, shared weights, etc. By integrating YOLOv8 and CvT, this method focuses on accurately classifying crowd density levels after identifying human presence in the waiting area (indoor). Evaluation metrics include mean Average Precision (mAP) for YOLOv8, and accuracy, precision, recall, and f1-score for CvT. This approach directly influences the management of HVAC systems.

Estimation of water quality in Korattur Lake, Chennai, India, using Bayesian optimization and machine learning

Assessing water quality becomes imperative to facilitate informed decision-making concerning the availability and accessibility of water resources in Korattur Lake, Chennai, India, which has been adversely affected by human actions. Although numerous state-of-the-art studies have made significant advancements in water quality classification, conventional methods for training machine learning model parameters still require substantial human and material resources. Hence, this study employs stochastic gradient descent (SGD), adaptive boosting (AdaBoosting), Perceptron, and artificial neural network algorithms to classify water quality categories as these well-established methods, combined with Bayesian optimization for hyperparameter tuning, provide a robust framework to demonstrate significant performance enhancements in water quality classification. The input features for model training from 2010 to 2019 comprise water parameters such as pH, phosphate, total dissolved solids (TDS), turbidity, nitrate, iron, chlorides, sodium, and chemical oxygen demand (COD). Bayesian optimization is employed to dynamically tune the hyperparameters of different machine learning algorithms and select the optimal algorithms with the best performance. Comparing the performance of different algorithms, AdaBoosting exhibits the highest performance in water quality level classification, as indicated by its superior accuracy (100%), precision (100%), recall (100%), and F1 score (100%). The top four important factors for water quality level classification are COD (0.684), phosphate (0.119), iron (0.112), and TDS (0.084). Additionally, variations or changes in phosphate levels are likely to coincide with similar variations in TDS levels.

Using Thermal Signature to Evaluate Heat Stress Levels in Laying Hens with a Machine-Learning-Based Classifier.

Infrared thermography has been investigated in recent studies to monitor body surface temperature and correlate it with animal welfare and performance factors. In this context, this study proposes the use of the thermal signature method as a feature extractor from the temperature matrix obtained from regions of the body surface of laying hens (face, eye, wattle, comb, leg, and foot) to enable the construction of a computational model for heat stress level classification. In an experiment conducted in climate-controlled chambers, 192 laying hens, 34 weeks old, from two different strains (Dekalb White and Dekalb Brown) were divided into groups and housed under conditions of heat stress (35 °C and 60% humidity) and thermal comfort (26 °C and 60% humidity). Weekly, individual thermal images of the hens were collected using a thermographic camera, along with their respective rectal temperatures. Surface temperatures of the six featherless image areas of the hens' bodies were cut out. Rectal temperature was used to label each infrared thermography data as "Danger" or "Normal", and five different classifier models (Random Forest, Random Tree, Multilayer Perceptron, K-Nearest Neighbors, and Logistic Regression) for rectal temperature class were generated using the respective thermal signatures. No differences between the strains were observed in the thermal signature of surface temperature and rectal temperature. It was evidenced that the rectal temperature and the thermal signature express heat stress and comfort conditions. The Random Forest model for the face area of the laying hen achieved the highest performance (89.0%). For the wattle area, a Random Forest model also demonstrated high performance (88.3%), indicating the significance of this area in strains where it is more developed. These findings validate the method of extracting characteristics from infrared thermography. When combined with machine learning, this method has proven promising for generating classifier models of thermal stress levels in laying hen production environments.

ARAware: Assisting Visually Impaired People with Real-Time Critical Moving Object Identification.

Autonomous outdoor moving objects like cars, motorcycles, bicycles, and pedestrians present different risks to the safety of Visually Impaired People (VIPs). Consequently, many camera-based VIP mobility assistive solutions have resulted. However, they fail to guarantee VIP safety in practice, i.e., they cannot effectively prevent collisions with more dangerous threats moving at higher speeds, namely, Critical Moving Objects (CMOs). This paper presents the first practical camera-based VIP mobility assistant scheme, ARAware, that effectively identifies CMOs in real-time to give the VIP more time to avoid danger through simultaneously addressing CMO identification, CMO risk level evaluation and classification, and prioritised CMO warning notification. Experimental results based on our real-world prototype demonstrate that ARAware accurately identifies CMOs (with 97.26% mAR and 88.20% mAP) in real-time (with a 32 fps processing speed for 30 fps incoming video). It precisely classifies CMOs according to their risk levels (with 100% mAR and 91.69% mAP), and warns in a timely manner about high-risk CMOs while effectively reducing false alarms by postponing the warning of low-risk CMOs. Compared to the closest state-of-the-art approach, DEEP-SEE, ARAware achieves significantly higher CMO identification accuracy (by 42.62% in mAR and 10.88% in mAP), with a 93% faster end-to-end processing speed.

Detection of human finger joints in ultrasound images: structure and optimization

Synovitis is the inflammation of a synovial membrane surrounding a joint. Its assessment is an important step in the diagnosis and treatment of rheumatoid arthritis. Joint detection is the first stage of an automated method of assessment of a degree of synovitis, from an Ultrasound (USG) image of a finger joint and its surrounding area. A joint detector consists of three parts: image preprocessing, feature extraction, and classification. Each part contains adjustable parameters that must be set experimentally to ensure the proper operation of the detector. Both the structure of a joint detector and a procedure for finding a near-optimal configuration of the adjustable parameters are described. The optimization process is based on two evaluation measures: Area Under the Receiver Operating Characteristic Curve (AUC) and False Positive Count (FPC). The optimization process decreases the number of pictures with multiple detections, which was the main point of works presented in this paper. This was achieved by increasing the number of components of the homogeneous mixed-SURF descriptor which has the greatest influence on the final result. Non-SURF descriptors achieve poorer classification results. Our research led to the creation of a better joint detector which could positively influence the final results of inflammation level classification.

Antimicrobial, antifungal, antioxidant activity and phytochemical investigation of fatty acids by GS/MS of raspberry (Rubus idaeus L.) shoot lipophilic extract

The aim of the study was to determine the content of fatty acids using GC-MS in the obtained extract, conduct a study of the antimicrobial, antifungal and antioxidant activities of Rubus idaeus shoot lipophilic extract. The quantification of fatty acids was accomplished through GC-MS, antioxidant activity was assessed by potentiometric method, antimicrobial and antifungal activity was determined by well method. The 18 compounds were identified by the GS-MS. The levulenic acid (64.47±0.08 mg/100 g), linoleic acid (8.50±0.08 mg/100 g) and linolenic acid (6.80±0.04 mg/100 g) dominated in the obtained lipophilic R. idaeus shoot extract. Bacillus subtilis (17.00±0.50 mm) was the most sensitive to lipophilic extract whereas P. vulgaris was the most resistant to the lipophilic extract. Moreover, Candida albicans was medium sensitive to lipophilic extract (13.50±0.50 mm). The antioxidant activity was 1.00 mmol equiv./mdry res, according to Maslov`s antioxidant level classification it has low level. The extract exhibited antimicrobial activity against all tested strains, with the most significant impact observed against B. subtilis. However, the obtained lipophilic extract showed a relatively low level of antioxidant activity. Consequently, the derivatives of fatty acids play a substantial role in the antimicrobial and antifungal effects, whereas their contribution to antioxidant activity appears to be limited.

Development of a Device and Algorithm Research for Akhal-Teke Activity Level Analysis

This study demonstrated that wearable devices can distinguish between different levels of horse activity, categorized into three types based on the horse’s gaits: low activity (standing), medium activity (walking), and high activity (trotting, cantering, and galloping). Current research in activity level classification predominantly relies on deep learning techniques, known for their effectiveness but also their demand for substantial data and computational resources. This study introduces a combined acceleration threshold behavior recognition method tailored for wearable hardware devices, enabling these devices to classify the activity levels of horses directly. The approach comprises three sequential phases: first, a combined acceleration interval counting method utilizing a non-linear segmentation strategy for preliminary classification; second, a statistical analysis of the variance among these segments, coupled with multi-level threshold processing; third, a method using variance-based proximity classification for recognition. The experimental results show that the initial stage achieved an accuracy of 87.55% using interval counting, the second stage reached 90.87% with variance analysis, and the third stage achieved 91.27% through variance-based proximity classification. When all three stages are combined, the classification accuracy improves to 92.74%. Extensive testing with the Xinjiang Wild Horse Group validated the feasibility of the proposed solution and demonstrated its practical applicability in real-world scenarios.

Psedo Spiritual Experience Religious and Cultural Tourism in West Java

This research focuses on the issue of the lack of spiritual experience gained by visitors to religious and cultural tourism in Indonesiatourism cases for the Al Jabbar Grand Mosque, the Mount Padang Site, and the Sagati Cultural Park in West Java.The aim of this research is to identify the spiritual level of visitors to religious and cultural tourism objects in West Java.This problem is solved based on theories of the spiritual experiences of religious and cultural tourism visitors with variablesworship, science, da'wah, tadabur, diversity, self-reflection, and the search for meaning.This research uses two approaches, namely qualitative and quantitative. A qualitative approach was carried out using the Maudhu'i Al Qur'an and Hadith analysis method to identify the main ideas for the development of religious and cultural tourism as a whole.based on visitors' spiritual experiences. The quantitative approach in this research uses a descriptive statistical analysis method of respondents' answers regarding the level of spiritual experience of visitors. The results of this research conclude the classification of the spiritual level of tourist objects through an analysis of the differences between the reality of respondents' satisfaction answers in the field and the theoretical conception of the spiritual experience that visitors should have. This research also provides recommendations for improving the quality of spiritual experiences at religious and cultural tourism objects.

Predictive Modeling of Combustible Water Content of Five Common Tree Species in Central Yunnan Province

The water content of forest combustible material is a key factor affecting forest fire and spread, and is the main reference factor for forest fire prediction and forecasting. The live branches, live leaves, dead branches and dead leaves of cap-dou oak, green oak, Yunnan poplar, Yunnan oil fir and silver wattle in central Yunnan were selected as the research objects, and the slope ( x1 ), slope direction ( x2 ), slope position ( x3 ), elevation ( x4 ), wind speed ( x5 ), wind direction ( x6 ), humidity ( x7 ), ground diameter ( x8 ), crown height ( x9 ), crown width ( x10), tree height (x11), depression (x12), and tree species (x13), and used correlation analysis to screen the main modeling factors, and constructed a prediction model of water content of combustible materials by multiple linear regression method. The results show that the correlation between the four factors of slope (x1), slope direction (x2), wind speed (x5) and humidity (x7) and the water content of combustible materials is significant, the water content decreases when the three influencing factors of slope (x1), slope direction (x2) and wind speed (x5) increase, and the water content is higher when the influencing factor of humidity (x7) increases, and the water content of combustible materials is constructed into a model of water content of 20 groups by these factors. Compare the predicted value of the water content prediction model with the actual value to analyze the error of the model, and the results show that the established model can be used to predict the water content. The errors were caused by rainfall during sampling, which absorbed water into the combustible materials and resulted in errors in the moisture content, and wind speed measurements, which resulted in inaccurate wind speed measurements. The established water content model can provide a basis for the classification of forest danger level and combustibility of combustible materials, and can provide help and reference for forest fire prevention and management.

Analisis Peran Stakeholder Terhadap Kesuksesan Proyek Konstruksi

Good management of the project is a prerequisite for achieving project objectives. Every construction company must be able to demonstrate the quality of its services so that it can excel in the competition. The problem in this study is whether the role of stakeholders affects the success of a construction project and how much influence it has on the success of the project. In this study the sample taken was the Department of Housing and Settlements of Banda Aceh City and obtained 6 respondents and PT Coplanar Daya Maxima obtained 4 respondents. Respondents in the purpose of this study were to find out what roles and how much influence stakeholders have on the success of construction projects. Data collection technique is distributing questionnaires to respondents. Data processing was carried out using validity and reliability tests then data analysis using descriptive analysis and product moment correlation, as well as multiple linear regression using SPSS version 25. This study consisted of independent variables (X) consisting of owners (X1), employees (X2), suppliers (X3), consumers (X4), competitors (X5) and business licenses (X6) and the selection variable (Y). The test results for the validity of the r-table value of 0.6319 meet the requirements for r-count > r-table. The results of the reliability test show that all independent variables (X) and stakeholder determinant variables (Y) are reliable. This is because it has met the minimum coefficient of Cronbach Alpha, which is greater than 0.6. The results of descriptive analysis have the lowest mean value found in the employee variable (X2) and business license (X6) of 4.100 and the highest mean value of the supplier variable (X3) of 4.475 which shows that the average respondent's answer is important, then all variables play an important role in project success. construction. The results of multiple linear regression obtained the following equation: Y = 24.018 + 0.412X1 – 0.002X2 – 0.759X3 – 0.526X4 + 0.427X5 – 0.701X6. The analysis shows that the dominant influence of the role of stakeholders on the success of the project is in the supplier variable (X3) with a regression coefficient value of -0.759 and the level of relationship between suppliers (X3) and stakeholders. The results of the classification of the relationship level on the value of the correlation coefficient based on the interval table of the correlation coefficient value and the strength of the relationship.

Novel metagenomics analysis of stony coral tissue loss disease.

Stony coral tissue loss disease (SCTLD) has devastated coral reefs off the coast of Florida and continues to spread throughout the Caribbean. Although a number of bacterial taxa have consistently been associated with SCTLD, no pathogen has been definitively implicated in the etiology of SCTLD. Previous studies have predominantly focused on the prokaryotic community through 16S rRNA sequencing of healthy and affected tissues. Here, we provide a different analytical approach by applying a bioinformatics pipeline to publicly available metagenomic sequencing samples of SCTLD lesions and healthy tissues from four stony coral species. To compensate for the lack of coral reference genomes, we used data from apparently healthy coral samples to approximate a host genome and healthy microbiome reference. These reads were then used as a reference to which we matched and removed reads from diseased lesion tissue samples, and the remaining reads associated only with disease lesions were taxonomically classified at the DNA and protein levels. For DNA classifications, we used a pathogen identification protocol originally designed to identify pathogens in human tissue samples, and for protein classifications, we used a fast protein sequence aligner. To assess the utility of our pipeline, a species-level analysis of a candidate genus, Vibrio, was used to demonstrate the pipeline's effectiveness. Our approach revealed both complementary and unique coral microbiome members compared to a prior metagenome analysis of the same dataset.

CNNFDNF: CONVOLUTIONAL NEURAL NETWORK AND CASCADE NEURO FUZZY NETWORK FOR SEVERITY LEVEL OF COVID-19 DETECTION USING CHEST X-RAY IMAGES

Chest X-ray (CXR) images are the most important imaging modality to monitor and detect several lung diseases and Coronavirus Disease 2019 (COVID-19) disease. This is the first imaging method in the detection of COVID-19. Since the limited availability of interpreted medical images, and the accuracy are the biggest challenges in the conventional methods. Hence, a novel framework is designed for the severity level of COVID-19 detection using CXR images named Convolutional Neural Network and Cascade Neuro-Fuzzy Network (CNNFDNF). Initially, the input CXR image is allowed to the preprocessing phase that is done utilizing the Kalman filter and Region of Interest (ROI) extraction. Then, the lung lobe segmentation is executed by Psi-Net. After that, the extraction of Convolutional Neural Network (CNN) features is conducted from the interested lobe region. Moreover, the extracted image features are subjected to COVID-19 detection, where it is achieved by Deep Generative Adversarial Network (Deep GAN) trained by Henry Gas Water Wave Optimization (HGWWO). Here, HGWWO is blended by Water Wave Optimization (WWO) and Henry Gas Solubility Optimization (HGSO). The detection phase is categorized into COVID and non-COVID. If the detected image is COVID, then the CXR images are given to the second level classification. Finally, it is accomplished by employing the proposed CNNFDNF to classify them into low, moderate, or high COVID-19 infection affected at lungs. The proposed CNNFDNF is the newly devised scheme by modifying the layers of CNN and cascaded neuro-fuzzy network (DNF) structure. The measures employed for this scheme namely, accuracy, sensitivity and specificity acquired are 93.0%, 94.1%, and 93.8%. The proposed method is essential for keeping the situation under control. Because, in the severe cases may have severe pneumonia, other organ failure, and possible death.

Water Level Classification for Early Flood Detection Using KNN Method

Floods occur when water levels exceed normal limits, causing rivers to overflow and inundate low-lying areas. Early warning systems for flood disasters are crucial to mitigate the damage caused, such as loss of life and property. A flood classification system can be developed by utilizing water level data from the Department of Water Resources to predict the likelihood of flooding using the K-Nearest Neighbors (KNN) algorithm. This study aims to determine the flood status classification based on water levels using the KNN method in the Ciliwung River. The research data were obtained from the DKI Jakarta open data site, consisting of 564 samples. The study evaluated K values ranging from 1 to 10. The average accuracy across all K scenarios was 99%, with the best K value being 1, which provided 100% accuracy, sensitivity, and specificity. These results indicate that the KNN method is effective in classifying flood status based on water level data, making it a reliable tool for early warning systems. This system is expected to help reduce the negative impacts of floods by providing accurate and timely information to the public and authorities. This research makes a significant contribution to the development of disaster mitigation technology, particularly in flood risk management in urban areas.

Isolation forest-voting fusion-multioutput: A stroke risk classification method based on the multidimensional output of abnormal sample detection

Background and ObjectiveStroke has become a major disease threatening the health of people around the world. It has the characteristics of high incidence, high fatality, and a high recurrence rate. At this stage, problems such as poor recognition accuracy of stroke screening based on electronic medical records and insufficient recognition of stroke risk levels exist. These problems occur because of the systematic errors of medical equipment and the characteristics of the collectors during the process of electronic medical record collection. Errors can also occur due to misreporting or underreporting by the collection personnel and the strong subjectivity of the evaluation indicators. MethodsThis paper proposes an isolation forest-voting fusion-multioutput algorithm model. First, the screening data are collected for numerical processing and normalization. The composite feature score index of this paper is used to analyze the importance of risk factors, and then, the isolation forest is used. The algorithm detects abnormal samples, uses the voting fusion algorithm proposed in this article to perform decision fusion prediction classification, and outputs multidimensional (risk factor importance score, abnormal sample label, risk level classification, and stroke prediction) results that can be used as auxiliary decision information by doctors and medical staff. ResultsThe isolation forest-voting fusion-multioutput algorithm proposed in this article has five categories (zero risk, low risk, high risk, ischemic stroke (TIA), and hemorrhagic stroke (HE)). The average accuracy rate of stroke prediction reached 79.59 %. ConclusionsThe isolation forest-voting fusion-multioutput algorithm model proposed in this paper can not only accurately identify the various categories of stroke risk levels and stroke prediction but can also output multidimensional auxiliary decision-making information to help medical staff make decisions, thereby greatly improving the screening efficiency.

Taoism-Net: A Fruit Tree Segmentation Model Based on Minimalism Design for UAV Camera

The development of precision agriculture requires unmanned aerial vehicles (UAVs) to collect diverse data, such as RGB images, 3D point clouds, and hyperspectral images. Recently, convolutional networks have made remarkable progress in downstream visual tasks, while often disregarding the trade-off between accuracy and speed in UAV-based segmentation tasks. The study aims to provide further valuable insights using an efficient model named Taoism-Net. The findings include the following: (1) Prescription maps in agricultural UAVs requires pixel-level precise segmentation, with many focusing solely on accuracy at the expense of real-time processing capabilities, being incapable of satisfying the expectations of practical tasks. (2) Taoism-Net is a refreshingly segmented model, overcoming the challenges of complexity in deep learning, based on minimalist design, which is used to generate prescription maps through pixel level classification mapping of geodetic coordinates (the lychee tree aerial dataset in Guangdong is used for experiments). (3) Compared with mainstream lightweight models or mature segmentation algorithms, Taoism-Net achieves significant improvements, including an improvement of at least 4.8% in mIoU, and manifested a superior performance in the accuracy–latency curve. (4) “The greatest truths are concise” is a saying widely spread by ancient Taoism, indicating that the most fundamental approach is reflected through the utmost minimalism; moreover, Taoism-Net expects to a build bridge between academic research and industrial deployment, for example, UAVs in precision agriculture.

Diabetic prediction and classification of risk level using ODDTADC method in big data analytics

Diabetic prediction and classification of risk level using ODDTADC method in big data analytics

Deep-learning model for background parenchymal enhancement classification in contrast-enhanced mammography

Background. Breast background parenchymal enhancement (BPE) is correlated with the risk of breast cancer. BPE level is currently assessed by radiologists in contrast-enhanced mammography (CEM) using 4 classes: minimal, mild, moderate and marked, as described in breast imaging reporting and data system (BI-RADS). However, BPE classification remains subject to intra- and inter-reader variability. Fully automated methods to assess BPE level have already been developed in breast contrast-enhanced MRI (CE-MRI) and have been shown to provide accurate and repeatable BPE level classification. However, to our knowledge, no BPE level classification tool is available in the literature for CEM. Materials and methods. A BPE level classification tool based on deep learning has been trained and optimized on 7012 CEM image pairs (low-energy and recombined images) and evaluated on a dataset of 1013 image pairs. The impact of image resolution, backbone architecture and loss function were analyzed, as well as the influence of lesion presence and type on BPE assessment. The evaluation of the model performance was conducted using different metrics including 4-class balanced accuracy and mean absolute error. The results of the optimized model for a binary classification: minimal/mild versus moderate/marked, were also investigated. Results. The optimized model achieved a 4-class balanced accuracy of 71.5% (95% CI: 71.2–71.9) with 98.8% of classification errors between adjacent classes. For binary classification, the accuracy reached 93.0%. A slight decrease in model accuracy is observed in the presence of lesions, but it is not statistically significant, suggesting that our model is robust to the presence of lesions in the image for a classification task. Visual assessment also confirms that the model is more affected by non-mass enhancements than by mass-like enhancements. Conclusion. The proposed BPE classification tool for CEM achieves similar results than what is published in the literature for CE-MRI.

PDPHE: Personal Data Protection for Trans-Border Transmission Based on Homomorphic Encryption

In the digital age, data transmission has become a key component of globalization and international cooperation. However, it faces several challenges in protecting the privacy and security of data, such as the risk of information disclosure on third-party platforms. Moreover, there are few solutions for personal data protection in cross-border transmission scenarios due to the difficulty of handling sensitive information between different countries and regions. In this paper, we propose an approach, personal data protection based on homomorphic encryption (PDPHE), to creatively apply the privacy computing technology homomorphic encryption (HE) to cross-border personal data protection. Specifically, PDPHE reconstructs the classical full homomorphic encryption (FHE) algorithm, DGHV, by adding support for multi-bit encryption and security level classification to ensure consistency with current data protection regulations. Then, PDPHE applies the reconstructed algorithm to the novel cross-border data protection scenario. To evaluate PDPHE in actual cross-border data transfer scenarios, we construct a prototype model based on PDPHE and manually construct a data corpus called PDPBench. Our evaluation results on PDPBench demonstrate that PDPHE cannot only effectively solve privacy protection issues in cross-border data transmission but also promote international data exchange and cooperation, bringing significant improvements for personal data protection during cross-border data sharing.

METHODOLOGIES FOR MANAGING THE TREEFALL RISK OF URBAN TREES

Recently, concern about the risk of falling trees, triggered by several fatal accidents involving falls in urban environments, has increased the interest of managers around the world in developing effective assessment techniques for accident prevention. The objective of this study was to compare different methods of visual assessment of the risk of falling urban trees in Spathodea campanulata individuals. To do so, 43 individuals were analyzed using five visual risk analysis methodologies. Data collection was conducted between December/2020 and January/2021 using a form created in Google Forms. The risk level classification was standardized into high, medium and low for all methodologies, and statistical tests were performed to verify the similarity between them. The time taken for analysis was timed and tree DBH dimensions were measured to verify the correlation with the degree of risk. It was observed that 41.8% of the trees were included in the high-risk class by all methodologies, with methodology II showing the lowest variation between individuals. Statistically, methodology I differed from all methodologies, while II was only the same as IV, and the others were similar. AV is the fastest and I takes the longest. No high correlation was found between time spent and degree of risk of falling trees or DBH.

Earthquake damage detection and level classification method for wooden houses based on convolutional neural networks and onsite photos

AbstractThe results of earthquake damage certification (EDC) surveys are the basis of support measures for improving the lives of disaster victims. To address issues such as a limited workforce to perform EDC surveys and difficulties in judging the level of damage, a damage detection and level classification method for wooden houses using multiple convolutional neural network models is proposed. The proposed method, including detection, filtering, and classification models, was trained and validated based on photographs collected from EDC surveys in Uki City, Kumamoto Prefecture. Then, a software system, which deployed these models, was developed for the onsite EDC surveyors to detect damages shown in the photographs of the surveyed house and classify damage levels. The test results based on 32 target buildings indicate that the detection model achieved high recall in detecting damage. Moreover, the redundant detected regions can be precisely filtered by the filtering model. Finally, the classification model achieved relatively high overall accuracy in classifying the damage level.