Face is the most dominant part of human body and a lot of information can be extracted from facial features. In this study deep learning models are used for classifying gender using facial images of male and female and applying the same model to classify the born gender of people who have gone through medical procedures to change their appearances and sometimes gender. The images were trained by several models. The performance of the models when tested on the images of subjects without alteration scored above 92%. The same models when tested on the transgender dataset were not able to reach the same success. The best score was 25% which indicates that the model is not able to identify transgenders born gender.
Gender Classification, Image Recognition, Transgender, Deep Learning, Convolutional neural Networks, Transfer Learning
Table 3: Precision, Recall and F1 Score for the Base Model with Dropout and Augmentation on Test Sets
Table 5: Average Precision, Recall and F1 Score for the Built Model and Transfer Learning Models on Transgender Test Set
