Dog Filters using Facial Landmark Detection

In this project, we apply dog filters to human images using Facial Landmark Detection. This project was made as an academic project for 6th semester in my B.E. at NSIT.

Description

In recent years, there has been a growing trend of mobile apps that allow addition of various filters like dog filters and flower crowns, on selfies. Even though there’s a divide on people’s opinion on these, but achieving this is quite interesting if regarded as a computer vision & programming challenge.

In this project, we make a tool to apply filters on human face images by facial landmark detection using Convolutional Neural Networks (CNNs).

Approach

1. Detect human faces in images using OpenCV.
2. Apply facial landmark detection to identify required points.
3. Apply the filter on the image using multiple operations.

(Model creation has been shown in the named python notebook and filter application is present in the file main.py)

Web-App

There is also a web-app which has been made using Flask. It uses Python library DLib to find various facial landmarks. To run the web-app,

1. Clone the repository.
2. cd "WebApp using DLib"/
3. Run python main.py
4. Open localhost in your browser.

Sample results

Made by:

Shashwat Uttam (356/CO/15)