The topic addressed in this research study concerns the combination of the Hough Transform with convolutional neural networks to improve pattern recognition in a collection of images. We propose a neural network model that takes as input a collection of images. In order to be able to compare the results obtained, the collection is processed by the Extended Hough Transform on the one hand, and on the other hand, it has not undergone any processing by the Extended Standard Hough Transform. The model proposed in our approach is composed of a set of convolutional layers and a fully connected layer. For this study, we used a dataset containing a total of 10,200 images. The experimental results obtained with our model give an accuracy of 70.00% with the dataset treated with the Extended Hough Transform and 66.67% with the other dataset. It can also detect images containing lines. In view of the experiments carried out, we have seen that the size of the learning base and the material resources are key factors in obtaining better results. Hough Transforms help to improve the accuracy of the convolutional neural network.