RESUMEN
This research explores machine learning algorithms for reservoir inflow prediction, including long short-term memory (LSTM), random forest (RF), and metaheuristic-optimized models. The impact of feature engineering techniques such as discrete wavelet transform (DWT) and XGBoost feature selection is investigated. LSTM shows promise, with LSTM-XGBoost exhibiting strong generalization from 179.81 m3/s RMSE (root mean square error) in training to 49.42 m3/s in testing. The RF-XGBoost and models incorporating DWT, like LSTM-DWT and RF-DWT, also perform well, underscoring the significance of feature engineering. Comparisons illustrate enhancements with DWT: LSTM and RF reduce training and testing RMSE substantially when using DWT. Metaheuristic models like MLP-ABC and LSSVR-PSO benefit from DWT as well, with the LSSVR-PSO-DWT model demonstrating excellent predictive accuracy, showing 133.97 m3/s RMSE in training and 47.08 m3/s RMSE in testing. This model synergistically combines LSSVR, PSO, and DWT, emerging as the top performers by effectively capturing intricate reservoir inflow patterns.
Asunto(s)
Algoritmos , Aprendizaje Automático , Modelos TeóricosRESUMEN
Water scarcity in arid regions poses significant livelihood challenges and necessitates proactive measures such as rainwater harvesting (RWH) systems. This study focuses on identifying RWH sites in Dera Ghazi Khan (DG Khan) district, which recently experienced severe water shortages. Given the difficulty of large-scale ground surveys, satellite remote sensing data and Geographic Information System (GIS) techniques were utilized. The Analytic Hierarchy Process (AHP) approach was employed for site selection, considering various criteria, including land use/land cover, precipitation, geological features, slope, and drainage. Landsat 8 OLI imagery, GPM satellite precipitation data, soil maps, and SRTM DEM were key inputs. Integrating these data layers in GIS facilitated the production of an RWH potential map for the region. The study identified 9 RWH check dams, 12 farm ponds, and 17 percolation tanks as suitable for mitigating water scarcity, particularly for irrigation and livestock consumption during dry periods. The research region was classified into four RWH zones based on suitability, with 9% deemed Very Good, 33% Good, 53% Poor, and 5% Very Poor for RWH projects. The generated suitability map is a valuable tool for hydrologists, decision-makers, and stakeholders in identifying RWH potential in arid regions, thereby ensuring water reliability, efficiency, and socio-economic considerations.
Asunto(s)
Sistemas de Información Geográfica , Lluvia , Pakistán , Abastecimiento de Agua , Monitoreo del Ambiente/métodosRESUMEN
An experimental setup was developed for simulating the field conditions to determine the force and power required for cutting cumin crops in dynamic conditions. The effect of cutter bar speeds, forward speeds, and blade type on cutting force and power requirement for cutting cumin were also studied. Experiments were carried out at three levels: cutter bar speeds, forward speeds, and blade type. The results showed that all the factors significantly affected cutting force. The cutting force followed a decreasing trend with the increase in cutter bar speed. Whereas it followed an increasing trend with the increase in forward speed. The maximum cutting force for all three blades was observed at a cutter bar speed of 2.00 strokes.s-1 and forward speed of 0.46 m.s-1. The idle power and actual power required for cutting the cumin crop were also determined based on the cutting force. The results obtained were validated by the power drawn from the power source while operating the cutter bar blades. The R2 values for Blade-B1, Blade-B2, and Blade-B3 were 0.90, 0.82, and 0.88, respectively. The cutting force was primarily affected by the cutter bar speed, resulting in PCR values of 74.20%, 82.32%, and 81.75% for Blade-B1, Blade-B2, and Blade-B3, respectively, followed by the forward speed, which also had an impact on PCR values of 16.60%, 15.27%, and 18.25% for Blade-B1, Blade-B2, and Blade-B3, respectively. The cutting force for Blade-B1, Blade-B2, and Blade-B3 varied from 15.96 to 58.97 N, 21.08 to 76.64 N, and 30.22 to 85.31, respectively, for the selected range of cutter bar speed and forward speed. Blade-B1 had 18 and 30% less power consumption than Blade-B2 and Blade-B3, respectively.
Asunto(s)
Productos Agrícolas , Productos Agrícolas/crecimiento & desarrollo , Nigella sativa , Producción de Cultivos/instrumentación , Producción de Cultivos/métodosRESUMEN
This article has been withdrawn at the request of the Editor-in-Chief.
RESUMEN
Dengue is one of Pakistan's major health concerns. In this study, we aimed to advance our understanding of the levels of knowledge, attitudes, and practices (KAPs) in Pakistan's Dengue Fever (DF) hotspots. Initially, at-risk communities were systematically identified via a well-known spatial modeling technique, named, Kernel Density Estimation, which was later targeted for a household-based cross-sectional survey of KAPs. To collect data on sociodemographic and KAPs, random sampling was utilized (n = 385, 5 % margin of error). Later, the association of different demographics (characteristics), knowledge, and attitude factors-potentially related to poor preventive practices was assessed using bivariate (individual) and multivariable (model) logistic regression analyses. Most respondents (>90 %) identified fever as a sign of DF; headache (73.8 %), joint pain (64.4 %), muscular pain (50.9 %), pain behind the eyes (41.8 %), bleeding (34.3 %), and skin rash (36.1 %) were identified relatively less. Regression results showed significant associations of poor knowledge/attitude with poor preventive practices; dengue vector (odds ratio [OR] = 3.733, 95 % confidence interval [CI ] = 2.377-5.861; P < 0.001), DF symptoms (OR = 3.088, 95 % CI = 1.949-4.894; P < 0.001), dengue transmission (OR = 1.933, 95 % CI = 1.265-2.956; P = 0.002), and attitude (OR = 3.813, 95 % CI = 1.548-9.395; P = 0.004). Moreover, education level was stronger in bivariate analysis and the strongest independent factor of poor preventive practices in multivariable analysis (illiterate: adjusted OR = 6.833, 95 % CI = 2.979-15.672; P < 0.001) and primary education (adjusted OR = 4.046, 95 % CI = 1.997-8.199; P < 0.001). This situation highlights knowledge gaps within urban communities, particularly in understanding dengue transmission and signs/symptoms. The level of education in urban communities also plays a substantial role in dengue control, as observed in this study, where poor preventive practices were more prevalent among illiterate and less educated respondents.
RESUMEN
The current investigation examines the effectiveness of various approaches in predicting the soil texture class (clay, silt, and sand contents) of the Rawalpindi district, Punjab province, Pakistan. The employed techniques included artificial neural networks (ANNs), kriging, co-kriging, and inverse distance weighting (IDW). A total of 44 soil specimens from depths of 10-15 cm were gathered, and then the hydrometer method was adopted to measure their texture. The map of soil grain sets was formulated in the ArcGIS environment, utilizing distinct interpolation approaches. The MATLAB software was used to evaluate soil texture. The gradient fraction, latitude and longitude, elevation, and soil texture fragments of points were proposed to an ANN. Several statistical values, such as correlation coefficient (R), geometric mean error ratios (GMER), and root mean square error (RMSE), were utilized to evaluate the precision of the intended techniques. In assessing grain size and spatial dissemination of clay, silt, and sand, the effectiveness and precision of ANN were superior compared to kriging, co-kriging, and inverse distance weighting. Still, less than a 50% correlation was observed using the ANN. In this examination, the IDW had inferior precision compared to the other approaches. The results demonstrated that the practices produced acceptable results and can be used for future research. Soil texture is among the most central variables that can manipulate agriculture plans. The prepared maps exhibiting the soil texture groups are imperative for crop yield and pastoral scheduling.
Asunto(s)
Arena , Suelo , Arcilla , Monitoreo del Ambiente/métodos , AgriculturaRESUMEN
Satellite remote sensing is widely being used by the researchers and geospatial scientists due to its free data access for land observation and agricultural activities monitoring. The world is suffering from food shortages due to the dramatic increase in population and climate change. Various crop genotypes can survive in harsh climatic conditions and give more production with less disease infection. Remote sensing can play an essential role in crop genotype identification using computer vision. In many studies, different objects, crops, and land cover classification is done successfully, while crop genotypes classification is still a gray area. Despite the importance of genotype identification for production planning, a significant method has yet to be developed to detect the genotypes varieties of crop yield using multispectral radiometer data. In this study, three genotypes of wheat crop (Aas-'2011', 'Miraj-'08', and 'Punjnad-1) fields are prepared for the investigation of multispectral radio meter band properties. Temporal data (every 15 days from the height of 10 feet covering 5 feet in the circle in one scan) is collected using an efficient multispectral Radio Meter (MSR5 five bands). Two hundred yield samples of each wheat genotype are acquired and manually labeled accordingly for the training of supervised machine learning models. To find the strength of features (five bands), Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Nonlinear Discernment Analysis (NDA) are performed besides the machine learning models of the Extra Tree Classifier (ETC), Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), Logistic Regression (LR), k Nearest Neighbor (KNN) and Artificial Neural Network (ANN) with detailed of configuration settings. ANN and random forest algorithm have achieved approximately maximum accuracy of 97% and 96% on the test dataset. It is recommended that digital policymakers from the agriculture department can use ANN and RF to identify the different genotypes at farmer's fields and research centers. These findings can be used for precision identification and management of the crop specific genotypes for optimized resource use efficiency.
Asunto(s)
Aprendizaje Automático , Triticum , Triticum/genética , Redes Neurales de la Computación , Modelos Logísticos , AgriculturaRESUMEN
This study aims to investigate the change in heavy metal concentration and evaluate pollution intensity using Sentinel-2 data. Sixty samples collected from the surface soil in the area were used to determine the concentration of lead, copper, and zinc using atomic absorption spectroscopy. Then, the step-by-step regression method was used in ArcGIS software to determine the relationship between the concentration of heavy metals and the ranking of the influential spectral bands of Sentinel-2 to monitor heavy metals in the relevant sampling points. According to the results, lead monitoring was effective through the blue channel, the ratio of green to near infrared-IV channels, and the ratio of short-wave infrared-III to near infrared-II channels. At the same time, copper was monitored through reflectance values in the red channel, the ratios of green to near infrared-IV channels, and the ratio of short-wave infrared-III to near infrared-II channels. The blue channel and the ratio of green to near infrared-IV channels the ratio of near infrared-II to near infrared-IV channels were efficient for zinc monitoring. Pollution Load Indices (PLI) and Geographical Accumulation Index (Igeo) were calculated to classify the contaminated soils of the region. The efficiency of each relationship obtained was evaluated using the root mean square error (RMSE) and Pearson's correlation coefficient (R). In summary, the copper, lead, and zinc equations had RMSE values of 1.8, 2.5, and 1.60 mg/kg, respectively. The Pearson correlation coefficients (R) for copper, lead, and zinc were 0.80, 0.76, and 0.72, respectively, which indicated good agreement between measured and estimated values.
RESUMEN
Many Bacillus species are essential antibacterial agents, but their antibiosis potential still needs to be elucidated to its full extent. Here, we isolated a soil bacterium, BP9, which has significant antibiosis activity against fungal and bacterial pathogens. BP9 improved the growth of wheat seedlings via active colonization and demonstrated effective biofilm and swarming activity. BP9 sequenced genome contains 4282 genes with a mean G-C content of 45.94% of the whole genome. A single copy concatenated 802 core genes of 28 genomes, and their calculated average nucleotide identity (ANI) discriminated the strain BP9 from Bacillus licheniformis and classified it as Bacillus paralicheniformis. Furthermore, a comparative pan-genome analysis of 40 B. paralicheniformis strains suggested that the genetic repertoire of BP9 belongs to open-type genome species. A comparative analysis of a pan-genome dataset using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Cluster of Orthologous Gene groups (COG) revealed the diversity of secondary metabolic pathways, where BP9 distinguishes itself by exhibiting a greater prevalence of loci associated with the metabolism and transportation of organic and inorganic substances, carbohydrate and amino acid for effective inhabitation in diverse environments. The primary secondary metabolites and their genes involved in synthesizing bacillibactin, fencing, bacitracin, and lantibiotics were identified as acquired through a recent Horizontal gene transfer (HGT) event, which contributes to a significant part of the strain`s antimicrobial potential. Finally, we report some genes essential for plant-host interaction identified in BP9, which reduce spore germination and virulence of multiple fungal and bacterial species. The effective colonization, diverse predicted metabolic pathways and secondary metabolites (antibiotics) suggest testing the suitability of strain BP9 as a potential bio-preparation in agricultural fields.
RESUMEN
Rainwater harvesting (RWH) is an essential technique for enhancing agricultural development, particularly in regions facing water scarcity or unreliable rainfall patterns. Water shortage, however, is one of the key causes of low crop production especially in mountainous regions like the Khyber Pakhtunkhwa province where most rainwater is lost by runoff. Therefore, rainwater harvesting could be a suitable to make better use of runoff and increase crop production. The study focuses on selecting suitable rainwater harvesting sites in District Karak to enhance agriculture by utilizing multi-influence factor (MIF) and fuzzy overlay techniques. We considered seven factors, i.e., land use land cover (LULC), slope, geology, soil, rainfall, lineament, drainage density, to create a ranking system to understand its application in site selection analysis. The results were combined into one overlay process to produce a rainwater harvesting suitability map. The weighted overlay analysis of the MIF model results reveals that 167.96 km2 area has a very high potential for rainwater harvesting, 874.17 km2 has a high potential, 1182.92 km2 has a moderate and 354.50 km2 has a poor potential for rainwater harvesting. The fuzzy overlay analysis revealed that 257.53 km2 has a very high potential for rainwater harvesting, 896.56 km2 area is classified as high, 1018.30 km2 moderate, and 407.7 km2 has poor potential for rainwater harvesting. The findings of this research work will help the policymakers and decision-makers construct various rainwater harvesting structures in the study area to overcome the water shortage problems.
Asunto(s)
Lluvia , Abastecimiento de Agua , Agricultura , Suelo , AguaRESUMEN
Land subsidence is considered a threat to developing cities and is triggered by several natural (geological and seismic) and human (mining, groundwater withdrawal, oil and gas extraction, constructions) factors. This research has gathered datasets consisting of 80 Sentinel-1A ascending and descending SLC images from July 2017 to July 2019. This dataset, concerning InSAR and PS-InSAR, is processed with SARPROZ software to determine the land subsidence in Gwadar City, Balochistan, Pakistan. Later, the maps were created with ArcGIS 10.8. Due to InSAR's limitations in measuring millimeter-scale surface deformation, Multi-Temporal InSAR techniques, like PS-InSAR, are introduced to provide better accuracy, consistency, and fewer errors of deformation analysis. This remote-based SAR technique is helpful in the Gwadar area; for researchers, city mobility is constrained and has become more restricted post-Covid-19. This technique requires multiple images acquired of the same place at different times for estimating surface deformation per year, along with surface uplifting and subsidence. The InSAR results showed maximum deformation in the Koh-i-Mehdi Mountain from 2017 to 2019. The PS-InSAR results showed subsidence up to -92 mm/year in ascending track and -66 mm/year in descending track in the area of Koh-i-Mehdi Mountain, and up to -48 mm/year in ascending track and -32 mm/year in descending track in the area of the deep seaport. From our experimental results, a high subsidence rate has been found in the newly evolving Gwadar City. This city is very beneficial to the country's economic development because of its deep-sea port, developed by the China-Pakistan Economic Corridor (CPEC). The research is associated with a detailed analysis of Gwadar City, identifying the areas with significant subsidence, and enlisting the possible causes that are needed to be resolved before further developments. Our findings are helpful to urban development and disaster monitoring as the city is being promoted as the next significant deep seaport with the start of CPEC.
RESUMEN
The present study is designed to monitor the spatio-temporal changes in forest cover using Remote Sensing (RS) and Geographic Information system (GIS) techniques from 1990 to 2017. Landsat data from 1990 (Thematic mapper [TM]), 2000 and 2010 (Enhanced Thematic Mapper [ETM+]), and 2013 to 2017 (Operational Land Imager/Thermal Infrared Sensor [OLI/TIRS]) were classified into the classes termed snow, water, barren land, built-up area, forest, and vegetation. The method was built using multitemporal Landsat images and the machine learning techniques Support Vector Machine (SVM), Naive Bayes Tree (NBT) and Kernel Logistic Regression (KLR). According to the results, forest area was decreased from 19,360 km2 (26.0%) to 18,784 km2 (25.2%) from 1990 to 2010, while forest area was increased from 18,640 km2 (25.0%) to 26,765 km2 (35.9%) area from 2013 to 2017 due to "One billion tree Project". According to our findings, SVM performed better than KLR and NBT on all three accuracy metrics (recall, precision, and accuracy) and the F1 score was >0.89. The study demonstrated that concurrent reforestation in barren land areas improved methods of sustaining the forest and RS and GIS into everyday forestry organization practices in Khyber Pakhtun Khwa (KPK), Pakistan. The study results were beneficial, especially at the decision-making level for the local or provincial government of KPK and for understanding the global scenario for regional planning.
RESUMEN
For sustainable land cover planning, spatial land cover models are essential. Deforestation, loss of agriculture, and conversion of pasture land to urban and industrial uses are only some of the negative consequences of human kind's insatiable need for more land. Using remote sensing multi-temporal data, spatial criteria, and prediction models can effectively monitor these changes and plan for sustainable land use. This research aims to predict the land use and land cover (LULC) with cellular automata (CA) and Markov chain models. Landsat TM, ETM + , and OLI/TIRS data were used for mapping LULC distributions for the years 1990, 2006, and 2022. A CA-Markov chain was developed for simulating long-term landscape changes at 16-year time steps from 2022 to 2054. Analysis of urban sprawl was carried out by using the support vector machine (SVM). Through the CA-Markov chain analysis, we expect that built-up area will grow from 285.68 km2 (22.59%) to 383.54 km2 (30.34%) in 2022 and 2054, as inferred from the changes that occurred from 1990 to 2022. Therefore, substantial deforestation area reduction will result if existing tendencies in change continue despite sustainable development efforts. The findings of this research can inform land cover management strategies and assist local authorities in preparing for the present and the future. They can balance expanding the city and preserving its natural resources.
Asunto(s)
Autómata Celular , Conservación de los Recursos Naturales , Humanos , Cadenas de Markov , Monitoreo del Ambiente , Agricultura , Análisis Espacio-Temporal , UrbanizaciónRESUMEN
Urban sprawl, also widely known as urbanization, is one of the significant problems in the world. This research aims to assess and predict the urban growth and impact on land surface temperature (LST) of Lahore as well as land use and land cover (LULC) with a cellular automata Markov chain (CA-Markov chain). LULC and LST distributions were mapped using Landsat (5, 7, and 8) data from 1990, 2004, and 2018. Long-term changes to the landscape were simulated using a CA-Markov model at 14-year intervals from 2018 to 2046. Results indicate that the built-up area was increased from 342.54 (18.41%) to 720.31 (38.71%) km2. Meanwhile, barren land, water, and vegetation area was decreased from 728.63 (39.16%) to 544.83 (29.28%) km2, from 64.85 (3.49%) to 34.78 (1.87%) km2, and from 724.53 (38.94%) to 560.63 (30.13%) km2, respectively. In addition, urban index, a non-vegetation index, accurately predicted LST, showing the maximum correlation R2 = 0.87 with respect to retrieved LST. According to CA-Markov chain analysis, we can predict the growth of built-up area from 830.22 to 955.53 km2 between 2032 and 2046, based on the development from 1990 to 2018. As urban index as the predictor anticipated that the LST 20-23 °C and 24-27 °C, regions would all decline in coverage from 5.30 to 4.79% and 15.79 to 13.77% in 2032 and 2046, while the temperature 36-39 °C regions would all grow in coverage from 15.60 to 17.21% of the city. Our results indicate severe conditions, and the authorities should consider some strategies to mitigate this problem. These findings are significant for the planning and development division to ensure the long-term usage of land resources for urbanization expansion projects in the future.
Asunto(s)
Monitoreo del Ambiente , Tecnología de Sensores Remotos , Temperatura , Monitoreo del Ambiente/métodos , Urbanización , CiudadesRESUMEN
This research aims to assess the urban growth and impact on land surface temperature (LST) of Lahore, the second biggest city in Pakistan. In this research, various geographical information system (GIS) and remote sensing (RS) techniques (maximum likelihood classification (MLC)) LST, and different normalized satellite indices have been implemented to analyse the spatio-temporal trends of Lahore city; by using Landsat for 1990, 2004, and 2018. The development of integrated use of RS and GIS and combined cellular automata-Markov models has provided new means of assessing changes in land use and land cover and has enabled the projection of trajectories into the future. Results indicate that the built-up area and bare area increased from 15,541 (27%) to 23,024 km2 (40%) and 5756 km2 (10%) to 13,814 km2 (24%). Meanwhile, water area and vegetation were decreased from 2302 km2 (4%) to 1151 km2 (2%) and 33,961 km2 (59%) to 19,571 km2 (34%) respectively. Under this urbanization, the LST of the city was also got affected. In 1990, the mean LST of most of the area was between 14 and 28 â, which rose to 22-28 â in 2004 and 34 to 36 â in 2018. Because of the shift of vegetation and built-up land, the surface reflectance and roughness of each land use land cover (LULC) class are different. The analysis established a direct correlation between Normalized Difference Water Index (NDWI), Normalized Difference Vegetation Index (NDVI) with LST and an indirect correlation among Soil Adjusted Vegetation Index (SAVI), Normalized Difference Built-up Index (NDBI), and Built-up Index (BI) with LST. The results are important for the planning and development department since they may be used to guarantee the sustainable utilization of land resources for future urbanization expansion projects.
Asunto(s)
Autómata Celular , Monitoreo del Ambiente , Temperatura , Pakistán , Monitoreo del Ambiente/métodos , AguaRESUMEN
A divorce is a legal step taken by married people to end their marriage. It occurs after a couple decides to no longer live together as husband and wife. Globally, the divorce rate has more than doubled from 1970 until 2008, with divorces per 1,000 married people rising from 2.6 to 5.5. Divorce occurs at a rate of 16.9 per 1,000 married women. According to the experts, over half of all marriages ends in divorce or separation in the United States. A novel ensemble learning technique based on advanced machine learning algorithms is proposed in this study. The support vector machine (SVM), passive aggressive classifier, and neural network (MLP) are applied in the context of divorce prediction. A question-based dataset is created by the field specialist. The responses to the questions provide important information about whether a marriage is likely to turn into divorce in the future. The cross-validation is applied in 5 folds, and the performance results of the evaluation metrics are examined. The accuracy score is 100%, and Receiver Operating Characteristic (ROC) curve accuracy score, recall score, the precision score, and the F1 accuracy score are close to 97% confidently. Our findings examined the key indicators for divorce and the factors that are most significant when predicting the divorce.
Asunto(s)
Divorcio , Máquina de Vectores de Soporte , Países Desarrollados , Femenino , Humanos , Modelos Lineales , Redes Neurales de la Computación , Estados UnidosRESUMEN
The development of smart network infrastructure of the Internet of Things (IoT) faces the immense threat of sophisticated Distributed Denial-of-Services (DDoS) security attacks. The existing network security solutions of enterprise networks are significantly expensive and unscalable for IoT. The integration of recently developed Software Defined Networking (SDN) reduces a significant amount of computational overhead for IoT network devices and enables additional security measurements. At the prelude stage of SDN-enabled IoT network infrastructure, the sampling based security approach currently results in low accuracy and low DDoS attack detection. In this paper, we propose an Adaptive Machine Learning based SDN-enabled Distributed Denial-of-Services attacks Detection and Mitigation (AMLSDM) framework. The proposed AMLSDM framework develops an SDN-enabled security mechanism for IoT devices with the support of an adaptive machine learning classification model to achieve the successful detection and mitigation of DDoS attacks. The proposed framework utilizes machine learning algorithms in an adaptive multilayered feed-forwarding scheme to successfully detect the DDoS attacks by examining the static features of the inspected network traffic. In the proposed adaptive multilayered feed-forwarding framework, the first layer utilizes Support Vector Machine (SVM), Naive Bayes (NB), Random Forest (RF), k-Nearest Neighbor (kNN), and Logistic Regression (LR) classifiers to build a model for detecting DDoS attacks from the training and testing environment-specific datasets. The output of the first layer passes to an Ensemble Voting (EV) algorithm, which accumulates the performance of the first layer classifiers. In the third layer, the adaptive frameworks measures the real-time live network traffic to detect the DDoS attacks in the network traffic. The proposed framework utilizes a remote SDN controller to mitigate the detected DDoS attacks over Open Flow (OF) switches and reconfigures the network resources for legitimate network hosts. The experimental results show the better performance of the proposed framework as compared to existing state-of-the art solutions in terms of higher accuracy of DDoS detection and low false alarm rate.
RESUMEN
Plant health is the basis of agricultural development. Plant diseases are a major factor for crop losses in agriculture. Plant diseases are difficult to diagnose correctly, and the manual disease diagnosis process is time consuming. For this reason, it is highly desirable to automatically identify the diseases in strawberry plants to prevent loss of crop quality. Deep learning (DL) has recently gained popularity in image classification and identification due to its high accuracy and fast learning. In this research, deep learning models were used to identify the leaf scorch disease in strawberry plants. Four convolutional neural networks (SqueezeNet, EfficientNet-B3, VGG-16 and AlexNet) CNN models were trained and tested for the classification of healthy and leaf scorch disease infected plants. The performance accuracy of EfficientNet-B3 and VGG-16 was higher for the initial and severe stage of leaf scorch disease identification as compared to AlexNet and SqueezeNet. It was also observed that the severe disease (leaf scorch) stage was correctly classified more often than the initial stage of the disease. All the trained CNN models were integrated with a machine vision system for real-time image acquisition under two different lighting situations (natural and controlled) and identification of leaf scorch disease in strawberry plants. The field experiment results with controlled lightening arrangements, showed that the model EfficientNet-B3 achieved the highest classification accuracy, with 0.80 and 0.86 for initial and severe disease stages, respectively, in real-time. AlexNet achieved slightly lower validation accuracy (0.72, 0.79) in comparison with VGGNet and EfficientNet-B3. Experimental results stated that trained CNN models could be used in conjunction with variable rate agrochemical spraying systems, which will help farmers to reduce agrochemical use, crop input costs and environmental contamination.