Alzheimer’s disease (AD) is the most common type of late-stage dementia. Brain’s volume often decreases in AD, and this affects many functions. Algorithms for precise early AD diagnosis have been developed using machine learning (ML) approaches. Nevertheless, the classifiers’ clinical usefulness, interpretability, and generalisability to datasets and MRI procedures are still restricted. In this research, novel techniques in neuroimaging-based segmentation and classification for AD detection utilizing ML method are proposed. Input is collected as MRI brain images and processed for noise removal with normalization here. An active graph cut U-net C-means neural network was used to segment the processed image. A transfer convolutional squeeze net Bayesian regression model was used to classify the image. In the experimental analysis, detection accuracy, AUC, mean, mean average precision, recall and F-1 score are calculated for different MRI brain image datasets. Proposed technique’s mean average precision is 95%, recall 97%, detection accuracy is 98%, F1-score 94%, and AUC 96%. Furthermore, the results are compared with previous studies, which concluded that the proposed model performs better.