Age Invariant Face Recogntion Using Stacked Autoencoder Deep Neural Network

When it comes to computer vision, the most challenging and crucial field is Face recognition with latest technology system. There are plenty of research articles available, and had done lot of research in this domain to achieve the stability, but due to the timely Changes new challenges has been occurred with time and the research started towards to resolve the problem. As we have analyzed due to the certain changes of time and structure of face has reshaped with the effect of aging. So, various authors has discussed the face recognition without aging affect in plenty of articles, but all have some problems and low accuracy to identify the face. This research also describes face recognition with different face features like structure, shape, texture, size and many more. In this paper, we mainly considers two domains, face recognition and Age Invariant. The aim of this research is to include the face recognition work which does not achieved by previous researchers that does not discussed the significant methods, such as: Feature based, Appearance based and soft computing based method. To evaluate our technique, we used a large dataset known as Cross-Age Celebrity Dataset (CACD). This dataset contains more than 2000 Celebrities having 160,000 Images with age ranging from 16 to 62. We have used 1000 Images as input dataset for training and model evaluation. We proposed the Auto-encoder denoising deep neural Network, which extracts the 1000 features from image, and achieved 99.8% Accuracy.