Dental caries is a prevalent oral disease that may cause serious complications if not diagnosed early. Accurate detection is essential to avoid further injury and reduce pain for the patient. Conventional diagnostic techniques, such as visual inspection and X-ray interpretation with the naked eye, cause delays and misdiagnoses. As a solution to this, in this work, we introduce a deep learning method for automatic caries detection from dental X-rays. Our algorithms, such as convolutional neural networks (CNNs) and vision transformer (ViT), were tested on labeled radiographs, of which U-Net yielded a maximum accuracy of 99%, followed by DeepLabV3+(98%) and MobileNet (97%). This artificial intelligence-based process improves diagnostic quality, facilitates early intervention, and provides a proven tool for clinicians. It has the potential for real-world translation in mobile dental clinics and settings with limited resources, enhancing the outcome for the patient and eliminating the risk for late-stage complications.