COCO Dataset validator

COCO_validator.py

import json

coco_file = r'/content/sample.json'

with open(coco_file) as json_file:
  coco_data = json.load(json_file)
  
 def assertions(key, values, required_keys, unique_key=None):
  unique_key_id_mapper = {}
  for value in values:
    if unique_key is not None:
      unique_key_id_mapper[value['id']] = value[unique_key]
    for required_key in required_keys:
      assert required_key in value, "'{}' does not contain the required key '{}'".format(key, required_key)
  return unique_key_id_mapper

def annotation_assertions(key, annotations, image_map, category_map):
  required_keys = ['area', 'iscrowd', 'bbox', 'category_id', 'ignore', 'segmentation', 'image_id', 'id']
  assertions('annotations', coco_data['annotations'], required_keys, None)
  for annotation in annotations:
    assert len(annotation['bbox']) == 4, "'{}' key in 'annotations' does not match the expected format".format('bbox')
    assert annotation['category_id'] in category_map, "'{}' is not present in the 'categories' mapping".format('category_id')
    assert annotation['image_id'] in image_map, "'{}' is not present in the 'images' mapping".format('image_id')
    assert annotation['area'] == (annotation['bbox'][2] * annotation['bbox'][3]), "Mismatch of values in '{}' and '{}'".format('area', 'bbox')
    assert len(annotation['segmentation'][0]) == 8 or len(annotation['segmentation']) == 0, "'{}' must either be an empty list or contain a list of 8 values".format('segmentation')
    assert annotation['iscrowd'] == 0 or annotation['iscrowd'] == 1, "'{}' must either be 0 or 1. {} is invalid".format('iscrowd', annotation['iscrowd'])

def main():
  required_keys = ['images', 'type', 'annotations', 'categories']
  for required_key in required_keys:
    assert required_key in coco_data.keys(), "Required key '{}' not found in the COCO dataset".format(required_key)
    assert len(coco_data[required_key]) > 0, "Required key '{}' does not contain values".format(required_key)

  image_map = assertions('images', coco_data['images'], ["file_name", "height", "width", "id"], "file_name")
  category_map = assertions('categories', coco_data['categories'], ["id", "name", "supercategory"], "name")
  annotation_assertions('annotations', coco_data['annotations'], image_map, category_map)
  print('The dataset format is COCO!')

if __name__ == '__main__':
  main()

Does this only support type 1?
https://cocodataset.org/#format-data

PS there doesn't seem to be a type key.

Does this only support type 1? https://cocodataset.org/#format-data

PS there doesn't seem to be a type key.

Unfortunately yes.

It supports only type 1 (object detection). I'll try to add for other types when free.

you did not check "ignore" field as well as "id"

You need to certify if that the access to the segmentation list is safe to conclude de validation in a right way.

I've made a quick update to resolve this:

if len(annotation['segmentation']) > 0:
assert isinstance(annotation['segmentation'][0], list), "'{}' should be a list of lists or an empty list".format('segmentation')
assert len(annotation['segmentation'][0]) == 8, "'{}' should contain a list of 8 values".format('segmentation')
else:
assert len(annotation['segmentation']) == 0, "'{}' should be an empty list".format('segmentation')

Please share if im not seeing something. Tks btw :)