UAV stands for unmanned Aerial Vehicle, which can be as small as birds, regular drones or as big as private aircraft with no pilot on board. Since there is no one onboard UAV is remotely controlled. UAV is currently being used for a different purpose, examples are spy footages, sky view footage, reconnaissance, attacking roles, aerial surveillance, motion picture filmmaking, disaster rescue, parcel delivery, warehouse management, and other uses. Due to UAV multi-functionality and portability especially drones demand is growing faster, therefore, people need systems that work with the UAV (drones) to detect objects in real-time for military, safety reconnaissance, and surveillance. This paper review approaches to detect objects from camera view and from UAV scenes using machine learning algorithms. The growth of computer vision systems initiated better algorithms using huge training and testing datasets, faster GPU and CPU so that systems can achieve state of the art object detection system by training and classifying the data using machine learning approach. Since there are different orientation, background, occlusion in an image, object detection is not an easy task. The goal of object detection is to categorize images and video feeds from UAV into common categories.