Dental health should always be prioritized to maintain overall hygiene, as it contributes to a healthier life. However, tartar buildup and discoloration pose significant challenges, often leading to periodontal diseases if left undetected. This work analyzed the presence and severity of tartar, based on its spread to the tooth and discoloration. A dataset of 3,888 images was constructed using various augmentation techniques, and deep learning models were experimented with. The results revealed that DeepLab v3 outperformed other models in segmentation accuracy, achieving an average Dice score of 70.14%, an average IoU of 61.77%, a precision average of 71.68%, a recall of 86.98%, and a Hausdorff Distance average of 6.43 mm. Moreover, the model recorded an average dis-coloration severity of 62%. These findings suggest that while the model exhibits moderate effectiveness in detecting, segmenting, and assessing tartar discoloration, further enhancements in parameter optimization, training strategies, and processing techniques are necessary to improve overall accuracy and performance.