Tables are usually used to present the main points of a document so that readers can quickly understand the content of the document. This study proposed a novel deformable convolutional neural network model for table detection to identify and extract tables from electronic document images. The model can perform table detection and table structure recognition at the same time, and more effectively detect the location of tables and columns. The proposed model is evaluated using Marmot extended dataset and the experimental results show that the table detection cycle is reduced, the computation time is shortened, and the overall efficiency is improved. Compared with other studies, the proposed model has achieved better table detection results in terms of precision, recall, and F1-score.