Scropper has been removed and is no longer available in NStack!

(S)mart Cropper

New in 2.3.0

A feature to find a focal point in an image via machine learning. The focal point will indicate where in the image the important things are.

Use cases:

Auto crop image to many aspect ratios. Eg: in a CMS when saving new image


Scropper takes an image as input and outputs coordinates for the focal points baesd on high contrast, and a bounding box around any faces (if any) in the image.

  • Scropper only works with .jpeg or .png images for now.
  • When sending a GET request with a URL to an image, the path must be directly to a readable image, e.g, this path won't work, while this path will work.
  • The first call to Scropper will be slow (15-20s wait) while Lambda spins up.
  • Latency is a function of image size. Right now there are no limits. Expect approximately 1s latency for a 1MB image.


Scropper outputs two things:

1) The relative position of the focal point in the image represented by distance from top-left corner in percent on the x-axis and y-axis on the image. This will always be there.

2) A bounding box around faces. This will be empty if there are no detected faces in the image.