In the context of Computer Vision, Pedestrian Detection (PD) has been a challenging task in various domains, viz. Advanced Driver Assistance Systems (ADAS), Suspicious Activity Systems (SAS), Autonomous Vehicle (AV), etc. Computer Vision algorithm works with different visual contents like individual still images as well as videos, which is a collection of frames for detection. This article represents a comprehensive review of pedestrian detection on these imagers using various Deep Learning (DL) approaches. A considerable amount of recent literature has been surveyed and introspected to fetch research gap on pedestrian detection under various ML/DL algorithms used by other researchers. Researchers formulated many techniques that automatically predict the human body in images and videos. In their proposed technique, many different scenarios over there to acquire high accuracy. This paper deals with a complete review of some existing techniques and new approaches to pedestrian detection. Though the previous survey works endorsed on achieving equal or more than 90% accuracy but many of those have not given insight into real-time detection and small-scale detection.