Face Recognition has become a major problem in order to identify the theft over surveillance systems, especially the occluded Face recognition technology has gain more importance now a days. there are lot of challenges faced by existing surveillance systems available even though various research works has been carried out still many of the existing machine learning techniques fail in identifying the occluded faces. By employing deep learning-based technique, the performance of face recognition tasks has been greatly boosted. The majority of cutting-edge methods still struggle with the verification and discriminating of faces with occlusions. In light of this, this research proposes a unique convolutional neural network that was created specifically for comparing occluded and non-occluded faces for the same identity utilizing CNN and BiLSTM approaches. Based on the architecture of multiple network convolutional neural networks, it could learn both the shared and distinctive properties. The training and testing of the proposed convolutional neural network incorporated the recently disclosed joint loss function and the accompanying alternating minimization strategy. The proposed deep convolutional neural network approach outperforms the state-of-the-art face identification algorithms by 10-15% in terms of several performance characteristics, according to experimental results on the publically accessible datasets (LFW 99.73%, YTF 97.30%, and CACD 99.12%).

Occluded Face Recognition, Deep Neural Network Technique, CNN, BILSTM, Image Processing.