With the rapid growth of the Internet and technologies over the last couple of decades, the trend of cheating in the examination is changing and increasing exponentially. The new technologies provide new means of cheating, and students can learn about different ways to avoid getting caught from social media, causing an upsurge in the cheating trend. Data studied in this work has been collected from an academic institution where examinations are supposed to be held under the invigilation of an invigilator. A method for predicting cheating activities has been proposed in this study. This method monitors the physical activities performed by the students during an examination, detects any anomalies relevant to cheating, and classifies them into different categories. This method integrates resizing the images in the whole dataset and their conversion to grayscale images, feature extraction, selection, and classification of images. Features are extracted by applying feature descriptors such as Local binary patterns and texture features (SFTA and Gabor). Afterward, the fused features are selected by implementing the Principal Component Analysis method. Lastly, the selected features are classified using a support vector machine and Fine KNN. The proposed method uses the newly created dataset to evaluate its effectiveness. This approach has provided a promising score of accuracy 73.8%, sensitivity 69.7%, specificity 6.06%, and F-Score 71.76%. This method detects and classifies the cheating activities performed by the students during their exams. This study however can also be used to monitor physical activities performed by students on various occasions such as the representation of different cultures in sports week.