Transfer Learning using quantization Facial recognition through static images can be done through deep neural networks.However, with deep neural networks the problem of vanishing gradients and high variance is still a major issue. A new proposed Resnet50 quantized transfer learning model has been implemented to the dataset. The effects of pruning and distinctiveness on the trained model weights has been observed. Although, the pretrained Resnet50 model increased the accuracy from the baseline dataset predictions. The combination of quantized transfer learning and Resnet50 resulted in an overall accuracy of 54% on the test set. The described model would be enough to classify static images in raw collection and the application of distinctiveness could help us to realize our pattern weights efficiently. A link on this paper is as follows :- https://drive.google.com/file/d/1uRke0hMBFSPz8frDJMJO1vCWUXU6ex2_/view?usp=sharing