Developing a YOLO based Object Detection Application using OpenCV
Mannem Ponika, K Jahnavi, P. Sridhar, Kavuri Veena
Abstract
Computer Vision is a technology in which computers are trained using machine learning and deep learning algorithms to recognize and process objects from video/images in the same way that humans do. Since 2012, Convolutional Neural Network and its variations have experienced exceptional improvement, making object detection a crucial application of image processing. The main issue with CNN is that this consumes more time to instruct/train as it has to analyze two thousand regions per picture so for this task in real time it needs approximately 47 sec per picture. This article has employed the YOLO real-time object identification approach to train our machine learning model. YOLO is one of the best CNN representatives that defies new way for object recognition, which is simple and highly efficient. Unlike existing algorithms, YOLO looks at the picture entirely by predicting the bounding boxes and class probabilities for these boxes using neural networks. This results in a faster detection of the image than other algorithms.