Driver drowsiness is a critical factor in road safety, and developing accurate models for detecting it is essential. Transfer learning has been shown to be an effective technique for driver drowsiness detection, as it enables models to leverage large, pre-existing datasets. However, the optimization of hyperparameters in transfer learning models can be challenging, as it involves a large search space. The core purpose of this research is on presenting an approach to hyperparameter tuning in transfer learning for driving fatigue detection based on Bayesian optimization and Random search algorithms. We examine the efficiency of our approach on a publicly available dataset using transfer learning models with the MobileNetV2, Xception, and VGG19 architectures. We explore the impact of hyperparameters such as dropout rate, activation function, the number of units (the number of dense nodes), optimizer, and learning rate on the transfer learning models' overall performance. Our experiments show that our approach improves the performance of the transfer learning models, obtaining cutting-edge results on the dataset for all three architectures. We also compare the efficiency of Bayesian optimization and Random search algorithms in terms of their ability to find optimal hyperparameters and indicate that Bayesian optimization is more efficient in finding optimal hyperparameters than Random search. The results of our study provide insights into the importance of hyperparameter tuning for transfer learning-based driver drowsiness detection using different transfer learning models and can guide the selection of hyperparameters and models for future studies in this field. Our proposed approach can be applied to other transfer learning tasks, making it a valuable contribution to the field of ML.

To improve the model's efficiency, people use many different methods, including the Transfer Learning algorithm, to improve the efficiency of recognition and classification of image data. The study was carried out to combine optimization algorithms with the Transfer Learning model with MobileNet, MobileNetV2, InceptionV3, Xception, ResNet50V2, DenseNet201 models. Then, testing on rice disease data set with 13.186 images, background removed. The result obtained with high accuracy is the RMSprop algorithm, with an accuracy of 88% when combined with the Xception model, similar to the F1, Xception model, and ResNet50V2 score of 87% when combined with the Adam algorithm. This shows the effect of gradients on the Transition learning model. Research, evaluate and draw the optimal model to build a website to identify diseases on rice leaves, with the main functions including images and recording of disease identification points for better management purposes on diseased areas of rice.

Plant disease identification at an early stage plays a crucial role in ensuring efficient management of the diseases and crop protection. The occurrence of plant ailments can result in substantial reductions in both crop yield and quality, which may cause financial setbacks for farmers and lead to food shortages for consumers. Traditional methods of disease detection rely on visual observation, which can consume a significant amount of time, be a labor-intensive, and often be inaccurate. Automated disease detection systems, based on techniques for machine learning have the potential to greatly improve the precision and speed of disease detection. This article presents a model for classifying plant diseases that combines the output of two transfer learning models, EfficientNetB0 and MobileNetV2, to improve disease classification accuracy. The PlantVillage Dataset was used to train and test the model under consideration, which contains 54,305 photos of 38 different plant disease classes, achieving an accuracy rate of 99.77% in disease classification. The use of an ensemble of deep learning models in this study shows promising results, indicating that the technique can enhance the accuracy of plant disease classification. Besides, this study contributes to the development of accurate and reliable automated disease detection systems, thereby supporting sustainable agriculture and global food security.

An autonomous vehicle is a vehicle that can run autonomously using a control. There are two modern autonomous assistant systems that are proposed in this research. First, we introduce a real-time approach to detect lanes of the streets. Based on a series of multi-step image processing through input data from the camera, the vehicle's steering angle is estimated for lane keeping. Second, the steering control system ensures that autonomous vehicles can operate stably, and smoothly, and adapt to various road conditions. The steering controller consists of a PID controller and fuzzy logic control strategy to adjust the controller parameters. The simulation experiments by Gazebo simulator of the Robot Operating System (ROS) not only indicate that the vehicle can keep the lane safely, but also demonstrate that the proposed steering angle controller is more stable and adaptive than the conventional PID controller.

No abstract