Monkeypox is an important health problem. Rapid diagnosis of monkeypox skin lesions and emergency isolation when necessary is essential. Also, some skin lesions, such as melanoma, can be fatal and must be rapidly distinguished. However, in some cases, it is difficult to distinguish the lesions visually. Methods such as dermoscopy, high-resolution ultrasound imaging, etc. can be used for better observation. But these methods are often based on qualitative analysis, subjective and time-consuming. Therefore, in this study, a quantitative and objective classification tool has been developed to assist dermatologists and scientists. The proposed system classifies seven skin lesions, including monkeypox. A popular approach Vision Transformer and some popular deep learning convolutional networks have been trained with the transfer learning approach and all results have been compared. Then, the models that show the best accuracy score have been combined to make the final prediction using bagging-ensemble learning. The proposed ensemble-based system produced 81.91% Accuracy, 65.94% Jaccard, 87.16% Precision, 74.12% Recall, and 78.16% Fscore values. In terms of different criteria metrics, the system produced competitive or even better results than the literature.

Correct and rapid classification of brain tumor types is crucial for the patient's treatment plans. This study aims to create a deep learning‐based mobile application that leverages on‐device AI capabilities to classify brain tumors. For this reason, first, a series of preprocessing steps are applied to MR images. Then, convolutional neural network , ViT, and MobileViT models are trained for this task. Also, pretrained VGG16, ResNet152V2, InceptionV3, InceptionResNetV2, and MobileNetV2 models are retrained for the brain tumor classification task with the transfer learning method. Using the publicly available “Brain Tumor MRI Dataset,” the model performances are evaluated, and test results are compared. MobileViT shows the best performance in terms of balance between inference time and success rate. Thus, the TensorFlow model of MobileViT is converted to the TensorFlow Lite model and integrated into the mobile application. The mobile application is developed using the Flutter framework. The application has been evaluated on two different devices, and 298.98 and 317.50 ms average inference times have been observed. The proposed system shows that rapid and effective brain tumor classification can be performed by integrating deep learning into the mobile application. This system can assist experts in the decision‐making process.

Garbage classification is essential for various aspects like health, economy and environment. The awareness of the users' separation of garbage has not been gained yet. So, garbage classification is done manually in waste sorting facilities. This is time-consuming and a difficult task; thus, automated systems are needed for this area. With this motivation, a deep learning-based system that provides automated classification of garbage has been developed in this study. In the study, five-types waste were used. These are glass, plastic, battery, paper and metal waste. MobileNetv3, ShuffleNet, EfficientNetb2, SqueezeNet, GoogleNet and MobileViT models were retrained with the transfer learning approach to classify garbage images. According to the training and testing processes performed on the Garbage Dataset, EfficientNetb2 obtained the highest performance with 0.9768 accuracy, 0.9769 precision, 0.9768 recall, and 0.9768 F1-score values. To better demonstrate the benefits of transfer learning approach, EfficientNetb2 was also trained from scratch and almost 18% smaller accuracy score than transfer learning was obtained. Moreover, to evaluate system generalization ability, the trained EficientNetb2 model was tested on another dataset. 0.9700 accuracy, 0.9706 precision, 0.9700 recall, 0.9701 f1-score values were obtained in the second dataset. The high scores on the second dataset may be related to less complex data included in this dataset and high-system generalization capacity. The future target of the study may be to integrate the system into an embedded system. Therefore, light-weight models were specially selected. These models use fewer parameters, run in less time, and require less processing load, which are requirements of the embedded systems. This system can be used in smart city projects, waste management systems, and even the development of robotic waste collection technologies. In addition to helping reduce environmental pollution, it can also contribute to reducing human labor and contributing to recycling.

Intelligent Transportation Systems are crucial to enhance traffic security and reduce traffic violations. Recent developments in artificial intelligence have also motivated researchers to work on these systems. In order to reduce the risk of road accidents, accurately determining the driving area is important. On the other hand, automated road segmentation is challenging because of different weather, light, road conditions, etc. Also, the roads’ design characteristics may be different and may include very small and large features. In this paper, a vision-based road segmentation system is presented. To apply this system, a CNN architecture based on consecutive triple filter size has been designed. The consecutive triple filter size approach avoids missing both small and large features. Proposed CNN architecture consists of an encoder and a decoder subnetwork. Preliminary experimental results achieved 95.84% IoU, 97.86% TPR, and 2.34% FPR indexes for road segmentation tasks on the CamVid database. This study presents a low-cost, reliable, and basic road segmentation system for intelligent vehicles. Also, it is a step toward a more complex system to alert the driver about dangerous situations. Owing to the proposed system, a successful road segmentation is possible using a simple camera.