Online Preview: https://kounelisagis.me/actors_matching/

A big number of people spend plenty of time watching Movies and Series. You can easily find and watch movies of your favorite actor. What if you could find movies with more than one actor of your choice?

Actors Matching is a web application which finds all the common Movies and Series based on the input actors.

• HTML, CSS, JQuery
• dlib face recognision library using Python
• SQLite
• themoviedb API
• imgur API

For the identification of the the input face a database is obviously needed. I used dlib library so my SQLite database looks like this:

Explanation
id: themoviedb id of the actor
name: the name of the actor
vec0 - vec127: a set of 128 face points

Actors can be inserted with two ways. Method #1 - Image: It can contain one or more actors. Their faces are recognised using dlib library running on a python server. Method #2 - Text: It can contain only the name of one actor.

Name search just makes a simple request to themoviedb and return the first result

• upload image(jpg and png tested and working) using JQuery and get the Imgur photo_id (imgur API)
• pass the photo_id to the python and return actor(s) (python server, SQLite)

• add each new actor to actors' list and find their movies and series (themoviedb API)
• update the UI with pictures, names and links
• delete image if the Image-Based Search was used

• Problem with photos which contain more than one actors.
• Enter press causes page refresh when the text method is used.
• Some movies have no photos. That makes them disappear.
• Upload of the same actor isn't detected.
• Image Uploading might take big amount of time when the internet connection is slow. Maybe resizing the image before the upload would help.
• App renders images of varying size and that can be visually bad in some cases. Fixed size images would be better.