Skip to content

Latest commit

 

History

History
58 lines (34 loc) · 2.48 KB

README.md

File metadata and controls

58 lines (34 loc) · 2.48 KB

Screenshot-2021-07-29-at-10-50-14-PM

Introduction

A web app which enables users to upload images, have them supersized or stylized with custom styles using deep learning models like neural style transfer & Super Resolution GAN, and optionally post them on the community gallery. Users can also see each other's posts on the gallery and like them.

This was a group project for the junior year course CSCI3100 Software Engineering taught by Prof. Michael Lyu at CUHK.

Tech stack: React, Flask, PyTorch, MongoDB, Docker

Architecture

Screenshot-2021-07-29-at-10-56-54-PM Screenshot-2021-07-29-at-10-45-11-PM

Frontend

login  stylize successful_stylize community

Deep Learning Models

Neural Style Transfer

Architecture

style_transfer

Example

Content Image

HK

Style Image

wave

Stylized Image

stylized

SRGAN

Architecture

srgan_architecture

Example

srgan_example

References

[1] Johnson, Justin, Alexandre Alahi, and Li Fei-Fei. "Perceptual losses for real-time style transfer and super-resolution." European conference on computer vision. Springer, Cham, 2016. [https://arxiv.org/abs/1603.08155]

[2] Ledig, Christian, et al. "Photo-realistic single image super-resolution using a generative adversarial network." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.