Skip to content
This repository has been archived by the owner on Dec 2, 2022. It is now read-only.
/ midv500 Public archive

Download and convert MIDV-500 annotations to COCO instance segmentation format

License

Notifications You must be signed in to change notification settings

fcakyon/midv500

Repository files navigation

Downloads PyPI version CI

Download and convert MIDV-500 datasets into COCO instance segmentation format

Automatically download/unzip MIDV-500 and MIDV-2019 datasets and convert the annotations into COCO instance segmentation format.

Then, dataset can be directly used in the training of Yolact, Detectron type of models.

MIDV-500 Datasets

MIDV-500 consists of 500 video clips for 50 different identity document types including 17 ID cards, 14 passports, 13 driving licences and 6 other identity documents of different countries with ground truth which allows to perform research in a wide scope of various document analysis problems. Additionally, MIDV-2019 dataset contains distorted and low light images in it.

teaser

You can find more detail on papers:

MIDV-500: A Dataset for Identity Documents Analysis and Recognition on Mobile Devices in Video Stream

MIDV-2019: Challenges of the modern mobile-based document OCR

Getting started

Installation

pip install midv500

Usage

  • Import package:
import midv500
  • Download and unzip desired version of the dataset:
# set directory for dataset to be downloaded
dataset_dir = 'midv500_data/'

# download and unzip the base midv500 dataset
dataset_name = "midv500"
midv500.download_dataset(dataset_dir, dataset_name)

# or download and unzip the midv2019 dataset that includes low light images
dataset_name = "midv2019"
midv500.download_dataset(dataset_dir, dataset_name)

# or download and unzip both midv500 and midv2019 datasets
dataset_name = "all"
midv500.download_dataset(dataset_dir, dataset_name)
  • Convert downloaded dataset to coco format:
# set directory for coco annotations to be saved
export_dir = 'midv500_data/'

# set the desired name of the coco file, coco file will be exported as "filename + '_coco.json'"
filename = 'midv500'

# convert midv500 annotations to coco format
midv500.convert_to_coco(dataset_dir, export_dir, filename)