Dietary intake recognition is a crucial aspect of monitoring the individual’s dietary requirements. Manual self-reporting methods are often tedious and suffer from human imperfections like recall bias and reporting error. Thus, automatic monitoring of food intake may result in accurate recognition. Food processing in the human body starts from the chewing of food in the mouth. As an important part of chewing the movements of the jaw, if accurately identified, can help to increase the recognition capability of the chewing system. This work aims to develop a non-invasive method of chewing detection by monitoring the mandible movements using machine learning (ML) approaches. An accelerometer embedded custom hardware configuration is used to acquire the relevant dataset. A robust Local Binary Pattern (LBP) approach is employed for comparing local data patterns. K-nearest neighbors (KNN), support vector machine (SVM), and random forest (RF) ML methods are used for recognition of mouth movement in vertical, horizontal, forward, and no-movements classes. The experimental results with average classification accuracy of 97% validate that the proposed LBP-based signal processing approach is effective in accurate recognition of mouth movement.