Skip to content

A machine learning approach to evolving visual content using computer vision, VQGAN, and CLIP

License

Notifications You must be signed in to change notification settings

grahamwaters/HowTimeFlies

Repository files navigation

The How Time Flies Project

A curious examination of the visual past with the computational tools of the present

1

Version 1.0.0

Created by Graham Waters

GitHub issues GitHub release (latest by date) GitHub last commit GitHub stars GitHub forks

ViewCount

Table of Contents

What is this?

This project is a collection of images of the same place at different times. The images are taken from google images and currently are not filtered for their historical accuracy or quality. In further implementations this will be remedied. It uses requests and google images to show how a query visually changes given a time parameter when searching google images.

Contributing

We welcome all contributions to this project as it is an evolving tool. as we move towards releasing our first release we are looking to maintain black formatting and abide by flake8 as well. to contribute please fork this repository and clone it to your machine so that you can make your changes and then submit a pull request for us to review.

if you have any great ideas about how this project could be changed feel free to submit them as an issue.

The Process

The program prompts the user to give several inputs:

What are you looking for? you type: London, England

from what year at the earliest? you type: 1920

to what year at the latest? you type: 1980

How many images do you want? (80 is limit per page) you type: 20

what MUST be included in the search? you type: England

what would you like to leave out of the search? you type: (blank)

retrieving 60 years worth of photos related to London, England`
dataline:London, England "in the year  (years go here) "" England "-
this is 1200 images in 60 folders.

Then you should see something like this:

 "London, England" in "the year 1974" "England" imagesize:500x500
searching...
Downloading 20 images...
Running total of collected images:40
 52%|████████▋       | 31/60 [01:09<00:04,  1.05it/s]

How to use it

  1. Clone the repo
  2. Install the requirements
  3. Run python3 howitschanged.py
  4. The user inputs a query
  5. The program uses the requests library to get the html of the google images page, and uses BeautifulSoup to parse the html.

About

A machine learning approach to evolving visual content using computer vision, VQGAN, and CLIP

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published