Digital imaging and artificial intelligence (AI) have had a major influence in the health sector and show great potential as tools for cancer diagnosis. The objective of this analysis was to train a deep learning model and its validation for automatically classifying oral lesion images. Here, we introduced a deep learning model named RID_Net, which combines three CNN models: ResNet152V2, InceptionV3, and DenseNet201. Additionally, we have used the lips and tongue image dataset for the evaluation of the model’s performance. The classification gained an accuracy of 96% on an average in the dataset. The deep learning model development has been studied in detail for automatic classification of abnormal oral lesions from oral clinical images to achieve satisfactory performance. Moreover, future directions involve investigating the addition of trained layers to identify patterns which distinguish the potentially malignant and malignant lesions from the benign ones.