Basic face detection

35 points

Hit the problem endpoint, grab the image_url value. The link expires in 5 seconds.

In the linked image you'll see a 8x8 grid of tiles, some of which contain faces.

Your task is straightforward: detect which tiles have a face in them. The result should be a list of two-element lists indictating which tiles successfuly went through face detection.

Use zero based indexing for row and column. So in a 8x8 tile grid, top left image is [0, 0], bottom right will be [7, 7].

For example, if there's only one face-containing tile and it's in the second row, fifth column, the answer would be [[1, 4]].

Getting the problem set

GET /challenges/basic_face_detection/problem?access_token=...

Problem JSON format will be in the following format:

  • image_url: a one-time URL of the image you have to analyze

Submitting a solution

POST /challenges/basic_face_detection/solve?access_token=...

Solution JSON structure:

  • face_tiles: a list of [row, column] pairs (as two element lists) pointing to the tiles that have faces in them (e.g. [[1, 3], [4, 5]])

Remember to use integers as values, not strings.

Why this challenge?

Face detection is nowadays front-and-center in so many places. It still seems like magic to a lot of people, though - and it's really not! This challenge is your excuse for getting into this stuff.

By the way - in case you're wondering - a green apple is not a face.

You too can support h^ by buying me a coffee. Cheers!
Also, I'd love to hear from you. Whatever you got, drop me an email. My h^ profile is here.