Tagging and classifying facial images in cloud environments based on KNN using MapReduce

Abstract Complexities to the human face require tremendous computing power to recognize and distinguish faces correctly. Research in this field pertains to human–computer interaction and object recognition—including pose estimation and the recognition of faces, expressions, and gestures. With the advent of online social networks such as Twitter, LinkedIn and Facebook, the number of images being uploaded is rapidly increasing. However, there is a severe lack of applications that can make use of this data. The proliferation of image data has led to a demand for research into large image set analysis. Such research will become the key basis for competition, because standard tools and procedures are not designed to handle massive datasets. Moreover, images are complex multidimensional structures and require solid computing techniques for recognition. It is extremely urgent to develop a new robust and efficient platform. Fortunately, the Hadoop platform provides a powerful framework for computationally intensive distributed processing under a free license. In this paper, we provide a full solution to facial image tagging and classification in a cloud environment using Hadoop and KNN. Experimental results confirm that this system makes significant improvements in performance. Furthermore, the effectiveness of the system is evaluated by comparing recognition rates and processing times.