Object Detection and Action Recognition using Computer Vision

Abstract

One of the main issues with computer vision is the recognition of objects and actions. Deep learning has significantly changed how society uses artificial intelligence since it first emerged a few years ago. The study was primarily designed for monitoring and proctoring reasons. Action recognition is used to keep track of the subject's motions, and object detection is used to locate objects in the scene. This proposed model is built in Python using real-time computer vision frameworks like Open CV and YOLO. Moreover, Open CV is used in image processing and machine learning. Up to 80 different things can be recognized from the data set of objects provided. You Only Look Once(YOLO), Faster Recurrent Convolutional Neural Networks (RCNN), and Single Shot Detector (SSD). When performance is more important than accuracy, YOLO excels while Faster RCNN and SSD do better. This technique efficiently detects objects without degrading performance. There are few challenges faced during the study auch as Viewpoint Variation, Deformation, Occlusion, Illumination Condition, Cluttered, Intra-Class Variation.