Maintaining healthy teeth is important because teeth are risky to be affected by bad bacteria. Because unhealthy teeth will result in cavities. Oral and dental health is a problem that often occurs every year. Problems that often occur in the mouth are dental caries or cavities. U-net is a learning solution in assignments to quantification tasks that often occur such as detecting membranes and structuring details in image settings. It cannot enable the unlimited segmentation of large images that are altered by the overlaptile strategy. U-net deep learning method achieves good results in the segmentation of medical images. The u-net method is very important for overcoming problems in segmenting specific dental images in the dentin section. This greatly helps the dent inspection process quickly and minimizes errors due to manuals that can provide advice between doctors, and can produce positive results. The method of learning in the net achieves quite good results in the segmentation of medical images. The steps per age that are set in the training are 2000, 2500, and 3000. The process consists of processes that produce dentine mask prediction output. Determination of the results is divided into three categories, which are commensurate, sufficient, and less. For comparable results that still need to be improved for training data with parameter settings above 3000