Skip to content

AndyYu25/Hacklytics-2022-ArtificiallyIntelligent

Repository files navigation

Hacklytics-2022-ArtificiallyIntelligent

Team Project for Hacklytics 2022 to develop an autoblurrer using Google's Vision API and a DCGAN blurrer.

Files:

VideoBlurrer.py: a prototype application using opencv and the Vision API to automatically blur video footage shot from a webcam.
DataProcessing.ipynb: a notebook used to preprocess data for BlurGAN. Used the Vision API to separate images of multiple people (originally derived from the WIDER FACE dataset) into bounding box-sized images.
BlurGAN.ipynb: a generative adversarial network developed to blur faces to a higher degree than conventional blurring algorithms, like Gaussian blur. Consists of a generator built on PyTorch and a discriminator built using the Vision AI.
BlurGAN-earlydemo.png: A blur filter developed by BlurGAN on its second training epoch.
GaussianBlurDemo.png: A picture with the subject's face blurred using a Gaussian Blur.

Works Cited:

Yang, S., Luo, P., Loy, C., & Tang, X. (2016). WIDER FACE: A Face Detection Benchmark. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

About

Team Project for Hacklytics 2022 to develop an autoblurrer using Google's Vision API and a GAN blurrer

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors