In recent years, deep learning algorithm has been one of the most used method in machine learning. Success rate of the most popular machine learning problems has been increased by using it. In this work, we develop an eye detection method by using a deep neural network. The designed network, which is accepted as an input by Caffe, has 3 convolution layers and 3 max pooling layers. This model has been trained with 16K positive and 52K negative image patches. The last layer of the network is the classification layer which operates a softmax algorithm. The trained model has been tested with images, which were provided on FDDB and CACD datasets, and also compared with Haar eye detection algorithm. Recall value of the method is higher than the Haar algorithm on the two datasets. However, according to the precision rate, the Haar algorithm is successful on FDDB dataset.