Face Recognition Based on Cascade Classifier Using Deep Learning
Praveen Kumar Mannepalli, Devendra Singh Kushwaha, Sanjay Kalamdhar, Vishakha Nagrale, Vikram Rajpoot
Abstract
The human face is significant because it is used as a tool of identification in our daily lives. One kind of biometric identification is Face Recognition (FR), which works by memorizing and using a person's unique facial features to identify that person in the future. Numerous researchers have been interested in biometric facial recognition technology due to its widespread use. Because of the non-contact nature of the process, FR technology surpasses other biometric-based authentication methods like fingerprint, palm print, & iris recognition. One such potential use for facial recognition (FR) algorithms is in remote identification when no physical touch or interaction with the target individual is required. This study proposes a method for constructing an FR system with a learning strategy depending on the DCNN algorithm. The approach of Deep Learning (DL) is presented for learning & assessing all samples & new inputs in a scheme. The ORL and YALE face datasets are used in studies that compare the findings to those produced by the Deep CNN approach. The proposed DL method uses a systematic approach to learning from and assessing new inputs and samples. The proposed model Deep CNN obtained 96.04 percentage points (or a validation accuracy of 62.2%) over the ORL dataset after 100 iterations with a loss of 17.33 percent. DCNN achieves a 99.99% training accuracy and a 99.89% validation accuracy without any data loss after 100 epochs of training on the Extended Yale B dataset. The proposed DL method analyses and rates are existing data from the system and fresh face inputs. In addition, it gathers information from the facial features & enhances the D-CNN algorithm for facial classification. Utilizing CNN (Convolutional Neural Networks), the accomplishments in several contests are advancing and becoming the focus of study.