Arabic calligraphy recognition and classification are considered challenging tasks due to the difficulty that accompanies Arabic handwriting in artistic forms. Recognizing and classifying these styles are essential for applications in cultural heritage and digital archiving. The research conducted in this work aims to develop a robust and efficient model for classifying different types of Arabic calligraphy using deep learning techniques. A diverse dataset was collected, including Arabic calligraphy for four popular styles: Diwani, Kofi, Naskh, and Roqaa. The proposed approach utilizes a Convolutional Neural Network (CNN) architecture constructed using the Inception module to predict the calligraphy style used in the image text. The dataset was pre-processed using Canny edge detection to highlight the important features and reduce complexity. The model was trained and tested, achieving an accuracy of about 74%, with Diwani being the most accurately classified style with Recall of 90% and f1-score of 80%. These results highlight the potential of deep learning methods in automating the classification of Arabic calligraphy and suggest areas for future enhancement regarding the expansion of datasets and refining the used models.