Text-to-Face generation using Deep Learning. This project combines two of the recent architectures StackGAN and ProGAN for synthesizing faces from textual descriptions.
The project uses Face2Text dataset which contains 400 facial images and textual captions for each of them. The data can be obtained by contacting either the RIVAL group or the authors of the aforementioned paper.
Implementation (with some experimentation) of the paper titled VARIATIONAL DISCRIMINATOR BOTTLENECK: IMPROVING IMITATION LEARNING, INVERSE RL, AND GANS BY CONSTRAINING INFORMATION FLOW (arxiv -> https://arxiv.org/pdf/1810.00821.pdf )
Implementation uses the PyTorch framework.
A variant of the Self Attention GAN named: FAGAN (Full Attention GAN). The architecture of this gan contains the full attention layer as proposed in this project.
The project uses package named attn-gan-pytorch created by me, which is available at
python package for self-attention gan implemented as extension of PyTorch nn.Module.
Also includes generic layers for image based attention mechanism. Includes a Full-Attention layer as proposed by in another project of mine called FAGAN.
An information and membership form collection portal for the organisation named "South
Asian Association of Women Geoscientists" (SAAWG). Usage of Play framework for
Scala for micro-services development and AngularJS for front-end.