We’ve all encountered the struggle of choosing the right picture to upload to Instagram. Besides facing the difficulty of choosing the picture itself, you’ve got to make sure that your Instagram uploads have variety. Or, maybe you’re one of those people who has to have a certain uniform aesthetic. Oh—let’s also not forget the whole ordeal of selecting that one filter that’s just right or figuring out that perfect hashtag. Point being, choosing the right picture to Instagram is a difficult process, and we wanted to create a way to simply this task to be not only more user friendly, but also more accurate. Thus, LikeafAI was born.

What it does

Likefai works to determine what picture content garners a user the most 'likes' on Instagram. All that Likefai needs is the username of a public instagram account. Afterwards, Likefai applies Clarifai's computer vision algorithm to help the user determine what kinds of posts the user's followers are most inclined to 'like'.

How I built it

Likefai was written in Python. After the user inputted their public Instagram username, Likefai would parse through the HTML file of the Instgram profile page to get the image url's for the user's most recent twelve posts. A list of those twelve url's would be returned and those url's would be run through Clarifai's computer vision algorithm to return twenty keywords describing what was in the picture. Afterwards, Likefai would use the number of likes associated with each keyword in an algorithm to return a ranking of which pictures you should upload to get the most likes. When given your public Google Photos url, which has your pictures from your phone synced to it, Likefai would run through those pictures, get keywords, and then return the top photos.

Challenges I ran into

The biggest challenge for our app that kept us from finishing it during the hackathon was learning how to use Flask. Since we wrote our app in Python, Flask was the best option for backend. However, none of us had ever used Flask nor had much experience with backend work before, so we ran into a lot of obstacles while trying to implement it.

Accomplishments that I'm proud of

The algorithm that Nick Terrile wrote to recommend what pictures in your Google photos library would get the most likes, learning how to use Python, figuring out how to parse information, such as photo urls, from urls

What I learned

More than half of our team had never used Python before

What's next for Likefai

We hope to get our app actually working by finishing the backend. We plan to accomplish this using Flask.

Share this project: