README.md

metadata

license: cc-by-4.0
task_categories:
  - image-to-text
  - object-detection
language:
  - en
size_categories:
  - 1K<n<10K

COCO Val2014 - Image Captioning & Object Detection

This dataset contains COCO 2014 validation set with captions and object annotations.

Dataset Structure

• image_id: COCO image ID
• image: The image file
• input_prompt: Instruction prompt for the model
• gt_objects: List of ground truth object categories
• gt_captions: List of ground truth captions (5 per image)

Usage

from datasets import load_dataset

dataset = load_dataset("your-username/dataset-name")

Source

Based on COCO 2014 dataset: https://cocodataset.org/