Nowadays, the wheat plant has been considered a crucial source of protein, energy, and micronutrients for people. The motivation behind this study comes from how to increase the wheat crop growth and prevent wheat diseases as this plant plays a significant impact on food security all over the world. Wheat plant diseases can be divided into fungal, bacterial, viral, nematode, insect pests, physiological and genetic anomalies, and mineral and environmental stress. Digital images containing the wheat plant disease are collected from different public sources like Kaggle and GitHub. In this study, an adaptive deep-learning model is developed to classify and detect various types of wheat diseases collected digitally in an efficient accurate manner. The dataset is split into two sets: approximately 80% of the data ( 8,946 images) for the training set and 20% (2,259 images) for the validation set. The training set is composed of 1445, 1478, 1557, 1510, 1424, and 1532 images of healthy, leaf rust, powdery mildew, septoria, stem rust, and stripe rust while the validation set contains 357, 360, 404, 402, 353 and 383 images respectively. The suggested method achieved 97.47% validation accuracy on the training set of images and a testing accuracy of 98.42% on the testing set. This study offers a method of training for the classification and detection of wheat diseases using a mix of recently established pre-trained convolutional neural networks (CNN), DenseNet, ResNet, and EfficientNet integrated with the one-fit cycle policy. In comparison to the current state of the art, the proposed model is accurate and efficient.