Tasks

1st Challenge – Document Boundary Segmentation:

The objective of this challenge is to develop boundary detection algorithms for different kinds of documents. The teams should develop an algorithm capable of receiving an image (input data) of a document and returning a picture of the same dimension with the background in black pixels and the region of the document in white pixels.
Baseline paper: Neves Junior et al. [1].

Figure 1: Document Boundary Segmentation image sample.

2nd Challenge – Zone Text Segmentation:

This challenge encourages the development of algorithms for automatic text detection of identification documents. The teams should develop an algorithm capable of detecting patterns in the provided dataset; that is, to receive an image (input data) of a document (without a background), and return a picture of the same dimension with non-interest regions in black pixels and regions of interest (text regions) in white pixels.
Baseline paper: Neves Junior et al. [1].

Figure 2: Zone Text Segmentation image sample.

3rd Challenge – Signature Segmentation:

The objective of this challenge is to develop algorithms to perform the task of handwritten signature segmentation on identification documents. Given an image of some document (input data), the model or technique applied should return as a result an image containing only the pixels of the handwritten signatures, that is, an image with the same size of the input image with the handwritten signatures as foreground (white pixels), and the background (black pixels).
Baseline papers: Lopes Junior et al. [2]; P.G.S. Silva et al. [3].

Figure 3: Signature Segmentation image sample.