Dataset Overview

Dataset Description

• Dataset Name: COMPTECH2022 "WhoSigned?"
• Dataset Size: 1K < n < 10K
• Task Category: Image Classification
• License: cc-by-4.0

Dataset Details

• Features:

  • Image: Data type is image
  • Label: Data type is categorical label, containing two classes: forged and genuine

• Splits:

  • Training Set: Contains 6,171 samples, size 10,544,187.713 bytes

• Data Files:

  • Training Set Path: data/train-*

Dataset Content

• The dataset includes over 5,000 handwritten signature images and corresponding cropped images for distinguishing genuine and forged signatures.
• Each image contains about 10 handwritten signatures from the same user ID.
• Images are cropped using a segmentation neural network, each cropped image containing a single handwritten signature.
• User IDs can be defined from image file names.

Creation Support

• The dataset was created by Toloka.ai with support from COMPTECH2022.