KB-Ref Dataset
Created by Wang et al. at 2020, the KB-Ref Dataset is a referring expression comprehension dataset containing 43K expressions on 16K images. Different with other referring expression dataset, it requires that each referring expression must use at least one external knowledge (the information can not be got from the image)., in English language. Containing 16 in JSON file format.
Dataset Sources
Here you can download the KB-Ref dataset in JSON format.
Download KB-Ref dataset JSON files
Fine-tune with KB-Ref dataset
Metatext is a powerful no-code tool for train, tune and integrate custom NLP models
Paper
Read full original KB-Ref paper.
Classify and extract text 10x better and faster 🦾
Metatext helps you to classify and extract information from text and documents with customized language models with your data and expertise.