The morphological changes in the blood vessels may provide important information on the diagnosis of diseases. Therefore, the detection and segmentation of vascular structures turns into a significant problem. Many methods have been proposed in the literature, ranging from manually designed filters to learning-based algorithms. However, the histogram of oriented gradients, which is often used in computer vision applications, has never been used for vessel detection or segmentation. In this study, artificial neural networks were trained using the histogram of oriented gradients and the performance in the vessel detection was evaluated for each network. DRIVE data set is used for training and testing. As a conclusion, vessel detection was performed with sensitivity of 88.69% in the tests and proposed method outperforms some of the methods in the literature.