Vision loss due to illness can result from various medical conditions that affect the eyes. Advanced devices like OCT and ultra-widefield retinal cameras are expensive, making them less accessible in resource-limited settings. While eye image capture devices have transformed disease detection, they require expert analysis, and also variability between observers can lead to inconsistent diagnoses. Many devices do not seamlessly integrate with electronic health records (EHR), complicating workflow and data management. In the detection and diagnosis of eye disorders, machine learning (ML) has emerged as a vital technique. Deep learning algorithms based on individual eye diseases have been created to diagnose them. These models still have a number of drawbacks, including complexity of computation, low precision brought on by class imbalance, and increased time consumption. In order to support ophthalmologists, this study suggests an effective software-hardware framework with edge AI integration that uses a refined EfficientNetB0 model with spatial attention CNN block for rapid and accurate identification of eight types of eye illnesses using fundus images. While using pre-trained models and heatmaps is common, combining EfficientNet-B0 as a feature extractor with ImageNet-trained models for downstream classification introduces a novel hybrid pipeline. The architecture of the developed efficient net B0 model is optimized for the low-cost hardware and the work is directed to develop a smart, low latency, cost-effective device for use in underprivileged areas. Publicly accessible datasets from various sources, including 13,300 fundus and private hospital images, are used in the experimental work. The EfficientNetB0 model outperformed all other models with an accuracy of 96.2% for multi-label tasks, followed by Unet with 88% and DenseNet169 with 91%. The key aspects of the process are model optimization for edge devices, efficient preprocessing, and leveraging TensorFlow Lite’s performance features for real-time disease detection with an average inference time of 5 s per image on the Raspberry Pi 5 platform.

Predicting and modeling the quality of water is essential to guarantee that the water is safe to drink. The chlorine content in water needs to be monitored in real-time to provide a consistent supply of drinkable water. Additionally, potassium and chlorine have a major impact on how appealing the water is, as they are important components that influence taste and odor. Therefore, to evaluate the levels of chlorine and potassium, this work presents a multivariable linear regression approach backed by a hybrid feature extraction method. To bridge the gap between the filter and wrapper approaches, a hybrid approach is used to remove unnecessary information and reduce processing time and complexity. Here the quantitative parameters, in conjunction with categorical parameters, are instrumental in enabling accurate prediction of two water quality parameters. The two developed multi-level regression (MLR) models for the prediction of potassium and chloride are useful when factors affecting water parameters fluctuate at the site level as well as over larger spatial or temporal scales giving consumers a visual representation of how each parameter influences prediction. The converged model outperforms in comparison with other machine learning algorithms with an MAE of 7.42e-15 for potassium and 3.72e-14 for chloride.

The need for strong watermarking methods to safeguard IP and guarantee authenticity has increased due to the widespread use of digital images across many platforms. At the same time, there has been a meteoric rise in the need for visual quality improvement techniques for images. This work introduces a new method that combines watermarking, picture augmentation, and machine learning classification to provide security, visual integrity, and durability. Common assaults on traditional watermarking systems include cropping, filtering, and compression. On the other hand, our suggested method uses ML classifiers to adaptively embed watermarks in the frequency or spatial domains, making it more resistant to typical assaults. The embedding approach ensures resilience while minimising perceptual distortion by dynamically adapting to picture content features using a classification model trained on a broad dataset. Images that have been watermarked have their visual quality improved since image enhancement is a part of the watermarking framework. The suggested method finds target areas and selectively conducts improvement operations using feature extraction and analysis. This customised method reduces the effect of watermark insertion while preserving crucial picture characteristics.