This study evaluates the performance of various classifiers and pre-trained models for dental implant state classification using preprocessed radiography images with masks. A dataset of 511 periapical images, including 275 for Bicon, 70 for Bego, and 166 for ITI implants, was expanded to 5110 images using data augmentation techniques such as rotation, flipping, and scaling. Preprocessing included resizing, sharpening, noise reduction, CLAHE-based contrast enhancement, implant-specific masking, and normalization. Classifiers including Convolutional Neural Networks (CNN), Convolutional Support Vector Machine (CSVM), Convolutional Decision Tree (CDT), and Convolutional Random Forest (CRF) were employed. Pre-trained models such as VGG16, ResNet50, and Xception enhanced feature extraction. Model performance was assessed using accuracy, precision, recall, F1 score, and ROC AUC, with fivefold cross-validation ensuring robustness. CRF achieved the highest performance for ITI with Bego implants, with accuracy of 0.8966, precision of 0.9364, recall of 0.9253, F1 score of 0.9304, and ROC AUC of 0.9351. CNN delivered the best results for Bicon with Bego implants, achieving 0.9533 accuracy. Among pre-trained models, VGG16 with preprocessed data achieved superior results for Bicon vs. ITI classification, with 0.9865 accuracy and 0.9877 ROC AUC. Data augmentation and preprocessing significantly improved classifier performance. Preprocessing steps, coupled with data augmentation, enhanced classification performance, ensuring robustness across models. CRF and CNN were the top-performing classifiers, with VGG16 excelling among pre-trained models. These results highlight the importance of data augmentation and preprocessing in improving dental implant classification accuracy.

Access to high-resolution underwater images is crucial for the conservation and development of marine resources. Light scattering and light absorption are two fundamental issues in improving the quality of underwater images. Many of the captured images have severe degradation, which harms the systems and activities that rely on these images. To address this problem, we introduce an auxiliary network to enhance the contrast of underwater images. This network consists of three critical components. In the first step, a decoder network is used to recover gradient maps and enhance the brightness of the images. Then, we take the help of a brightness adjustment network to control the brightness of the hidden image, and finally, we use an adaptive contrast module to adjust the contrast. To improve the performance, we use a normalizer module to solve the problem of not paying attention to the increase in image contrast when increasing the brightness. Evaluation of the proposed method with public dataset images shows that our method can increase the resolution of underwater images. In addition, the proposed model can increase the resolution of images in complex images, low-light, and dark conditions.

Accurately forecasting the energy consumption of electric buses is important for fleet operation, charging planning, and service reliability. This study proposes an interpretable and uncertainty-aware machine learning framework for mission-level electric bus energy prediction under heterogeneous operating conditions. Using the public ZTBus dataset as a case study, the framework combines a global XGBoost model with cluster-specific models to examine variability across urban, mixed, and highway operating regimes. SHAP-based explainability is used to provide global feature attribution and mission-level local explanations, linking predictions to operationally meaningful factors such as speed, stop frequency, elevation, route distance, and environmental conditions. To support uncertainty-aware planning, bootstrap ensembles are used to characterize prediction dispersion, while conformal prediction intervals provide calibrated operational coverage. On temporally unseen data, the global XGBoost model achieves MAE ≈ 3.2 kWh and R² ≈ 0.90, while cluster-specific modeling reduces RMSE by up to approximately 28% in high-variability urban conditions. The framework shows moderate robustness under temporal distribution shift and recovers near-baseline performance after incremental retraining, while maintaining inference latency below 150 ms per prediction in the evaluated setting. Overall, the results suggest that combining heterogeneity-aware learning, explainability, and calibrated uncertainty estimation can support mission-level electric bus energy prediction within the evaluated ZTBus case study. Broader generalization should be confirmed through external validation across additional fleets, cities, vehicle types, and climatic conditions.