A Review on Fuzzy Face Recognition (FFR) using DCGAN
S. Nikkath Bushra, L. Javid Ali · 2021
AI based Facial recognition system has a number of applications in the area of biometric for authenticating a person, smart card identification, AI driven surveillance camera to identify criminals, health sectors, public security system to recognize suspected persons and as well as criminals in society and security mechanism in industry is also based on facial recognition system to identify their employees. Images used in Facial recognition system especially in real scenario have low resolution pictures taken under dark background with inadequate illumination, blurred images taken from low end cameras which has poor pixel quality. It is a cumbersome process to detect the face of a person exactly from inadequate information provided. To obtain precise output, huge amount of input data is required for training the neural network. This paper presents a comprehensive review on facial recognition system for improving improper, low quality and low light images by applying Artificial Intelligence based image and video analysis technique called Deep Convolutional Generative Adversarial Neural Network (DCGAN). DCGAN is commonly applied when the datasets are insufficient for training the neural network. The images to be detected under environment with poor illumination, due to climatic changes or due to capturing low resolution images using low quality cameras. Description regarding training and testing various datasets extracted from several freely available sources has been provided. The enactment of each model is considered and compared with other state-of-art models with respect to accuracy, complexity and execution time.