Identifying Moving Objects Using Background Subtraction, Optical Flow Method, Deep Neural Network for Camera in Motion Using PYNQ Architecture: Review of Recent Research Trends

The moving object recognition is a standout amongst the most required research subject which is blasting among the image processing specialists. Artificial intelligence is a hypothesis that is utilized to make master framework that can perform exercises with a similar knowledge that of people. As of late, the amazing capacity with the feature learning and exchange learning Binary Neural Network (BNN) includes developing interest inside the computer vision network. Moving object detection is the assignment of recognizing the physical development of a object in a given area. From most recent couple of years it has gotten much attraction because of applications like video observation, human motion analysis, robot navigation, event detection, traffic analysis and security. In this article we reviewed techniques to distinguish moving items dependent on optical flow and Background subtraction. The abnormality score is figured dependent on Background subtraction and relies upon contrast in pixel intensities between the current picture and the Background model. However, undertaking of detecting genuine shape of object in motion turns into a precarious because of different difficulties like dynamic scene changes, illumination varieties, presence of shadow, disguise and bootstrapping issue. To conquer these issues, scientists have proposed number of new methodologies. The outline of this article is to have a look at current advancement in observation frameworks and object detection. Their likelihood to get to irregular conduct and human object identification is the subject in different applications.