Background: Agriculture is one of the most crucial assets of any country, as it brings prosperity by alleviating poverty, food shortages, unemployment, and economic instability. The entire process of agriculture comprises many sectors, such as crop cultivation, water irrigation, the supply chain, and many more. During the cultivation process, the plant is exposed to many challenges, among which pesticide attacks and disease in the plant are the main threats. Diseases affect yield production, which affects the country’s economy. Over the past decade, there have been significant advancements in agriculture; nevertheless, a substantial portion of crop yields continues to be compromised by diseases and pests. Early detection and prevention are crucial for successful crop management. Methods: To address this, we propose a framework that utilizes state-of-the-art computer vision (CV) and artificial intelligence (AI) techniques, specifically deep learning (DL), for detecting healthy and unhealthy cotton plants. Our approach combines DL with feature extraction methods such as continuous wavelet transform (CWT) and fast Fourier transform (FFT). The detection process involved employing pre-trained models such as AlexNet, GoogLeNet, InceptionV3, and VGG-19. Implemented models performance was analysed based on metrics such as accuracy, precision, recall, F1-Score, and Confusion matrices. Moreover, the proposed framework employed ensemble learning framework which uses averaging method to fuse the classification score of individual DL model, thereby improving the overall classification accuracy. Results: During the training process, the framework achieved better performance when features extracted from CWT were used as inputs to the DL model compared to features extracted from FFT. Among the learning models, GoogleNet obtained a remarkable accuracy of 93.4% and a notable F1-score of 0.953 when trained on features extracted by CWT in comparison to FFT-extracted features. It was closely followed by AlexNet and InceptionV3 with an accuracy of 93.4% and 91.8% respectively. To further improve the classification accuracy, ensemble learning framework achieved 98.4% on the features extracted from CWT as compared to feature extracted from FFT. Conclusion: The results show that the features extracted as scalograms more accurately detect each plant condition using DL models, facilitating the early detection of diseases in cotton plants. This early detection leads to better yield and profit which positively affects the economy.

Skin diseases in melanin-rich skin often present diagnostic challenges due to the unique characteristics of darker skin tones, which can lead to misdiagnosis or delayed treatment. This disparity impacts millions within diverse communities, highlighting the need for accurate, AI-based diagnostic tools. In this paper, we investigated the performance of three machine learning methods -Support Vector Machines (SVMs), Random Forest (RF), and Decision Trees (DTs)-combined with state-of-the-art (SOTA) deep learning models, EfficientNet, MobileNetV2, and DenseNet121, for predicting skin conditions using dermoscopic images from the HAM10000 dataset. The features were extracted using the deep learning models, with the labels encoded numerically. To address the data imbalance, SMOTE and resampling techniques were applied. Additionally, Principal Component Analysis (PCA) was used for feature reduction, and fine-tuning was performed to optimize the models. The results demonstrated that RF with DenseNet121 achieved a superior accuracy of 98.32%, followed by SVM with MobileNetV2 at 98.08%, and Decision Tree with MobileNetV2 at 85.39%. The proposed methods overcome the SVM with the SOTA EfficientNet model, validating the robustness of the proposed approaches. Evaluation metrics such as accuracy, precision, recall, and F1-score were used to benchmark performance, showcasing the potential of these methods in advancing skin disease diagnostics for diverse populations.

Data security is one of the biggest concerns in the modern world due to advancements in technology, and cryptography ensures that the privacy, integrity, and authenticity of such information are safeguarded in today’s digitally connected world. In this article, we introduce a new technique for the construction of non-linear components in block ciphers. The proposed S-box generation process is a transformational procedure through which the elements of a finite field are mapped onto highly nonlinear permutations. This transformation is achieved through a series of algebraic and combinatorial operations. It involves group actions on some pairs of two Galois fields to create an initial S-box $P_{r}Sbox$, which induces a rich algebraic structure. The post S-box $P_{o}Sbox$, which is derived from heuristic group-based optimization, leads to high nonlinearity and other important cryptographic parameters. The proposed S-box demonstrates resilience against various attacks, making the system resistant to statistical vulnerabilities. The investigation reveals remarkable attributes, including a nonlinearity score of 112, an average Strict Avalanche Criterion score of 0.504, and LAP (Linear Approximation Probability) score of 0.062, surpassing well-established S-boxes that exhibit desired cryptographic properties. This novel methodology suggests an encouraging approach for enhancing the security framework of block ciphers. In addition, we also proposed a three-step image encryption technique comprising of Row Permutation, Bitwise XOR, and block-wise substitution using $P_{o}Sbox$. These operations contribute to adding more levels of randomness, which improves the dispersion across the cipher image and makes it equally intense. Therefore, we were able to establish that the approach works to mitigate against statistical and cryptanalytic attacks. The PSNR, UACI, MSE, NCC, AD, SC, MD, and NAE data comparisons with existing methods are also provided to prove the efficiency of the encryption algorithm.

Data security and privacy have become essential due to the increasingly advanced interconnectivity in today’s world, hence the reliance on cryptography. This paper introduces a new algorithm that uses a novel hybrid Tent–May chaotic map to generate pseudo-random numbers, as well as block encryption. We design a robust S-box by combining the Tent and May Maps, which yields a chaotic system with improved cryptographic properties. This S-box is a critical cryptographic primitive that significantly improves encryption security and leverages the strengths of both maps. The encryption process involves two key steps: block-wise substitution and permutation. First, we divide the image into $16\times 16$ blocks, then substitute each pixel with the $8-byte$ key and S-box. Next, we convert the encrypted image back into vector form, reorganize it using the permutation vector based on the subgroups of $S_{16}$, and finally return it to its original form. This approach greatly improves block cipher security when used, especially to protect medical images by guaranteeing their confidentiality and noninterference. Performance measures like PSNR, UACI, MSE, NCC, AD, SC, MD, and NAE prove how immune our method is to various cryptographic and statistical attacks, making it more accurate and more secure than the existing techniques.

Accurate air quality forecasting is important in pollution prevention and risk reduction. Effective short-term and long-term forecasting models are needed. This study investigated the need for a new model to forecast air pollution concentrations in three cities with distinct characteristics: a city with high industrial activity, a city with a high population density and urbanization, and an agricultural city. The air pollution data were collected using the Sentinel-5P satellite and Google Earth Engine to apply descriptive analysis and comparison of two years, 2022 and 2023. The studied cities were Al Riyadh (high population), Al Jubail (industrial), and Najran (agricultural) in Saudi Arabia. The selected pollutants were SO₂, NO₂, CO, O₃, and HCHO. In addition, this study investigated the variations observed in all the pollutants during the months of the year, the correlations between the contaminants, and the correlation between NO₂ and the meteorological data. Based on our findings, Al Jubail had the highest level of all the pollutants during the two years, except for NO₂, for which the highest level was observed in Al Riyadh, which has witnessed notable urbanization and development recently. Moreover, this study developed a forecasting model for the concentration of NO₂ based on weather data and the previous values of NO₂ using Long Short-Term Memory (LSTM) and Time2Vec. The modeling proved that any model that is trained on data collected from a specific city is not suitable for predicting the pollution level in another city and the level of another pollutant, as the three cities have different correlations with the pollutants and the weather data. The proposed model demonstrated a superior accuracy in predicting NO₂ concentrations compared to traditional LSTM models, effectively capturing temporal patterns and achieving minimal prediction errors, which contributes to ongoing efforts to understand the dynamics of air pollution based on cities’ characteristics and the period of the year.