README.md

metadata

dataset_info:
  features:
    - name: id
      dtype: string
    - name: url
      dtype: string
    - name: eng_caption
      dtype: string
    - name: score
      dtype: float64
    - name: kn_caption
      dtype: string
  splits:
    - name: test
      num_bytes: 5223531
      num_examples: 14906
    - name: train
      num_bytes: 258046154
      num_examples: 733604
  download_size: 156666204
  dataset_size: 263269685
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
      - split: train
        path: data/train-*
license: mit
task_categories:
  - visual-question-answering
language:
  - kn
  - en
size_categories:
  - 100K<n<1M

laion-coco dataset with captions translated to Kannada language. The dataset contains 733604 training and 14906 test samples. Images can be downloaded directly from Coco page.

Data Sample:

{'id': 'dde3bdc5-36b7-4340-b2ae-d9564c0d213a',
 'url': 'https://i.pinimg.com/236x/ca/84/a1/ca84a1d6f83c88c94452a94e320f024c--lens.jpg',
 'eng_caption': 'Black and white photograph of woman in hat leaning against tree.',
 'score': 5.8029,
 'kn_caption': 'ಮರದ ವಿರುದ್ಧ ಒರಗಿರುವ ಟೋಪಿ ಹೊಂದಿರುವ ಮಹಿಳೆಯ ಕಪ್ಪು ಮತ್ತು ಬಿಳಿ ಛಾಯಾಚಿತ್ರ.'}

Use with Datasets:

from datasets import load_dataset

ds = load_dataset("Kannada-LLM-Labs/Laion-Coco-Kn")