The simultaneous classification and grading of fruits are essential yet underexplored facets of computer vision in agricultural automation. This study proposes the application of same-domain transfer learning using the EfficientNetV2 architecture to facilitate multi-fruit classification and grading. Our dual-model framework initially employs EfficientNetV2 to distinguish between six fruit types—bananas, apples, oranges, pomegranates, limes, and guavas—within the FruitNet dataset. Subsequently, the learned parameters are transferred to a second model, which focuses on grading the quality of the fruits. To address the class imbalance in the dataset, we incorporate a combination of AugMix, CutMix, and MixUp, significantly improving model generalization. Our experiments demonstrate robust performance, with both classification and grading achieving an average test accuracy of 99%. These findings affirm the utility of same-domain transfer learning in enhancing grading accuracy using knowledge gained from classification tasks. The study shows promising potential for integrating this approach into machine vision systems to advance agricultural automation. Moving forward, this approach could be scaled to address broader cultivation challenges through the continued development of fine-grained visual analysis capabilities.
This research addresses the critical need for accurate solar energy forecasting in Saudi Arabia’s renewable energy transition. Leveraging the comprehensive Solar Resource Monitoring Stations dataset (2017-2021) from King Abdullah City for Atomic and Renewable Energy, this study presents the first extensive analysis of solar stations across Saudi Arabia with a systematic classification of solar stations into five distinct geographical groups, accounting for regional climate variations and solar radiation patterns. Using meteorological variables, including air conditions, wind patterns, relative humidity, and barometric pressure from 44 stations, we evaluated three machine learning models—Linear Regression (LR), Support Vector Regression (SVR), and Decision Tree (DT)—for predicting daily solar radiation. The results demonstrate that LR achieved superior performance with an RMSE of 139.71, significantly outperforming other models, particularly the Decision Tree model, which showed the highest error rate RMSE: 504.67. Through detailed regional analysis, identified northern and central regions exhibit consistently high Global Horizontal Irradiance (GHI), marking these areas as optimal locations for residential photovoltaic system deployment. These findings provide valuable insights for optimizing solar resource allocation and integration in Saudi Arabia’s power grid, facilitating the country’s transition to sustainable energy. This study’s results contribute to a better understanding solar resource distribution and variability, supporting informed decisionmaking for the Kingdom’s sustainable energy objectives.
Skin conditions influence people of all ages and genders and impose an enormous strain on worldwide public health. For efficient management and medical treatment, skin disorders must be accurately categorized. However, the conventional method of classifying skin conditions can be arbitrary and time-consuming, delaying diagnosis and treatment. In this research, we examine the application of traditional machine learning models and conventional image characteristics for the classification of skin cancer based on picture features. Specifically, we employ six feature extraction approaches, which we model using six classical classifiers. To evaluate our approach, we address skin cancer detection as both a seven-class problem and a two-class problem comprising 21 permutations of skin cancer instances. Our experimental results demonstrate that Random Forest achieves the highest performance, followed by Support Vector Machines. Additionally, our analysis reveals that the Edge Histogram and Fuzzy Opponent Histogram feature sets perform best in learning the skin cancer model. Our comprehensive evaluation of various models provides practitioners with valuable insights when selecting appropriate models for similar problems. Our findings demonstrate that acceptable detection performance can be achieved even with simple feature extraction and non-deep classifiers. We argue that classical features are not only easier and faster to extract than deep features but can also be combined with classical machine learning models to save time and valuable resources.
To minimize the spread of COVID-19, the Kingdom of Saudi Arabia (KSA) enforced a nationwide lockdown. We aimed to explore whether the manner in which Saudi patients with type 1 diabetes (T1D) manage their disease has changed during this unparalleled lockdown. An online survey exploring the effect of lockdown on T1D outcomes was distributed among T1D patients residing in KSA during lockdown. A total of 1010 patients responded to the survey. Around 40% reported communicating with their physicians during lockdown. Age, level of education, residence, previous visits to diabetes education clinics, last HbA1c value, and average monthly income were all significantly associated with communication with the treating physician (p = 0.008, p < 0.001, p < 0.001, p = 0.002, p < 0.001, and p < 0.001, respectively). Age, level of education, and average monthly income were significantly associated with experiencing severe hypoglycemia (p = 0.036, p = 0.03, and p < 0.001, respectively), while average monthly income and level of education were significantly associated with experiencing diabetic ketoacidosis (DKA) (p < 0.001 and p = 0.0039, respectively); during lockdown. Patients who communicated with their physicians reported lower rates of severe hypoglycemia compared to those who did not (25.2% vs 30.7%, respectively). Age and level of education were significantly associated with communication with the treating physician, and experiencing severe hypoglycemia and DKA; in patients with T1D during the lockdown period in Saudi Arabia. Keeping two-way virtual communication channels between physicians and their T1D patients should be encouraged.
At certain stages of the graphics pipeline, and most notably during compression for transmission and storage, triangle meshes may undergo a fixed-point arithmetic quantisation of their vertex coordinates. This paper presents the results of a psychophysical experiment, where discrimination thresholds between the original unquantised triangle meshes, and the quantised at various levels of quantisation versions of them, were estimated. The experiment had a two-alternative forced choice design. Our results show that the amount of geometric information of a mesh, as measured by its filesize after compression, correlates with the discrimination threshold. On the other hand, we did not find any correlation between the discrimination thresholds and the quality of the underlying meshing, as measured by the mean aspect ration of the mesh triangles.