Dhomne, Amit (2018) Age, Gender, Expression Recognition Through Face Using Secure Deep Learning. MTech thesis.
Restricted to Repository staff only
Facial attribute recognition including age, gender and expression has increase curiosity among computer vision and pattern recognition researchers to do research in this area from last few years.The main reason is the number of applications that work on this challenging area ranges from security control to person identification, to human-computer interaction. Facial Expression recognition has been used in various application actively such as avatar animation, neuromarketing and sociable robots.To identify particular facial expressions from various facial expression is not an easy task for machine learning methods because there are various way of showing expression. Even images of the same person in one expression can vary in brightness, background and position.
This is reason why facial expression recognition is also a challenging problem with age and gender identification from facial image. Availaibity of large labeled dataset and the advancement done in the design of convolutional neural networks, error rates have dropped significantly.However, this still remains a troublesome issue and existing commercial systems fall short when dealing with real-world scenarios. Convolutional neural system (CNN), a standout amongst the most commonly utilized deep learing methods, has been applied to number of application related to computer vision and pattern recognition tasks, and has achieved state-of-the-art performance.Three tasks are performed,namely age estimation and gender estimation from facial image and another task is to estimate expression from input face image.
To address the problems of age, gender and expression recognition in this work we propose a facial recognition system with facial attrbutes such as age, gender and expression that uses Convolutional Neural Networks. Data augmentation and different preprocessing steps were studied together with various Convolutional Neural Networks architectures. The data augmentation and pre-processing steps were used to help the network on the feature selection. Implementing Convolution Neural network(CNN) in Facial Identification Device(FID) helps to make system much more secure and error free. We have shown that the consistency factor and reliability has improved with the use of convolution neural network through the analysis and research.
|Item Type:||Thesis (MTech)|
|Uncontrolled Keywords:||CNN ; Deep learning; Face identification system; Neural network; VGGNet; Gender recognition ; Age recognition ; Expression recognition.|
|Subjects:||Engineering and Technology > Computer and Information Science > Networks|
|Divisions:||Engineering and Technology > Department of Computer Science Engineering|
|Deposited By:||IR Staff BPCL|
|Deposited On:||04 Apr 2019 20:08|
|Last Modified:||04 Apr 2019 20:08|
|Supervisor(s):||Sa, Pankaj Kumar|
Repository Staff Only: item control page