Abstract—Automated detection of human identity and gender offers several industrial applications in near future, such as monitoring, surveillance, commercial profiling and human computer interaction. In this paper, deep learning techniques have been used to investigate the problem of human identity and gender classification using hand images. First, pre-processing techniques have been applied to enhance the appearance of the hand images. The pre-processed image is passed through the convolution neural network to determine the gander. For identity detection, the network has been trained on the images for the determined gender for better recognition. To further enhance the result, the framework has been implemented using different optimizers and k fold cross-validation.  Experimental results have shown that highly effective performance is achieved in both the human identification and gender classification objectives. High average accuracy of 97.75% using the dorsal hand side for human identification and 96.79% has been obtained for gender classification using the palm hand side. Conclusively, the proposed method has achieved more accuracy than the previous methods both for identification and gender classification.

Keywords—human identification, gender classification, deep learning
 

Cite: Afaf Tareef, Hayat Al-Dmour, and Afnan Al-Sarayreh, "An Automated Deep Learning Framework for Human Identity and Gender Detection," Journal of Advances in Information Technology, Vol. 14, No. 1, pp. 94-101, February 2023.