Dataset Store

Full body portrait dataset

Full body portrait dataset (5 images per front/back/left/right side, 20 images per person)

Purchase Dataset

Tag

human

Image

Collecting and labeling of portrait datasets to train AI

Collection of image data to train AI with same-person detection algorithm

Image data for

same-person detection

Data collection and labeling were completed using the mobile version of Datumo’s crowd-sourcing platform, Cash Mission, in 2021.

Dataset specification

20 image files per object(person), minimum of 30*60pxl
8,000 persons*20 images (=160,000 image data)
Unstructured (JPG or PNG)
Size: 0MB~1TB

Process of annotation

*Person in different outfit is considered as different person

Data Collection

Cash Mission’s users(crowd-workers) participated in data collection and labeling by taking full body portrait photos according to categories such as [Standing upright-front/left/right/back] & [Walking-front/left/right/back].

1) There are five postures- [1Standing upright] and [4Walking].
2) The aspect ratio is [9:16]
3) Outfits and backgrounds for each set(20 images) are identical.
4) Each person(object) is wearing clothes(top&bottom or a dress) and shoes.
5) All faces have been blurred for privacy.

Portion of data were collected and labeled using Cash Mission, Datumo's crowd-sourcing platform.

Sample Data

Application

Person detection
Same-person detection