Deep learning models have revolutionized the field of Computer Vision by offering advanced techniques for interpreting visual information. Throughout the past years deep learning approaches and applications have been displayed to beat past cutting edge machine learning methods in countless fields with computer vision being considered one of the conspicuous cases. The specified review paper gives a concise outline of Convolutional Neural Networks (CNNs), Deep Boltzmann Machines (DBMs) and Deep Belief Networks (DBNs), Autoencoders and Extreme Learning, the major deep learning methods exploited in computer vision issues. A brief record of their set of experiences, structures, benefits, and constraints is given, trailed by a depiction of their applications in different computer vision errands, object detection, face recognition, action, and activity recognition. Finally, a compact overview is given of future headings in designing deep learning schemes and techniques for the issues included in computer vision.