You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Dec 2, 2022. It is now read-only.
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.
Download and unzip desired version of the dataset:
# set directory for dataset to be downloadeddataset_dir='midv500_data/'# download and unzip the base midv500 datasetdataset_name="midv500"midv500.download_dataset(dataset_dir, dataset_name)
# or download and unzip the midv2019 dataset that includes low light imagesdataset_name="midv2019"midv500.download_dataset(dataset_dir, dataset_name)
# or download and unzip both midv500 and midv2019 datasetsdataset_name="all"midv500.download_dataset(dataset_dir, dataset_name)
Convert downloaded dataset to coco format:
# set directory for coco annotations to be savedexport_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 formatmidv500.convert_to_coco(dataset_dir, export_dir, filename)
About
Download and convert MIDV-500 annotations to COCO instance segmentation format
