Transfer learning using AlexNet with convolutional neural network for face recognition

Yap, Yvonne (2020) Transfer learning using AlexNet with convolutional neural network for face recognition. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Transfer learning using AlexNet with convolutional neural network for face recognition.pdf - Submitted Version Download (2MB)

Abstract

This research is aimed to achieve high percentage accuracy and effectiveness face recognition system using small datasets. In recent years, the Deep Learning method used for image classification purposes. Convolution Neural Network is one of the Deep Learning approaches and has demonstrated excellent performance in many fields, including image recognition of a large amount of training data (such as ImageNet). In fact, hardware limitations and insufficient training datasets are the challenges of getting high performance. Therefore, I proposed a Deep Transfer Learning method using AlexNet pre-trained CNN to improve the performance of the face recognition system even for a smaller number of images. In this system, the Transfer Learning method will do fine-tuning on the last layer of the AlexNet CNN model for new classification tasks. AlexNet pre-trained CNN model by replacing the last layer for new classification tasks. Besides, the data augmentation technique is proposed to minimize the overfitting problem during Deep Transfer Learning training and to improve accuracy. The results proved the improvement in overfitting and in performance after using the data augmentation technique. All the experiments were tested on UTeMFD, GTFD, and CASIA-Face V5 small datasets. As a result, the proposed system achieved a high accuracy as 100% on UTeMFD, 96.67% on GTFD,and 95.60% on CASIA-Face V5 in less than 0.05 seconds of recognition time. Lastly, the results of this proposed system are better compared to other researchers using GTFD and datasets but different methods.

Actions (login required)

View Item