---
license: apache-2.0
language:
- en
tags:
- image recognition
task_categories:
- zero-shot-classification
size_categories:
- 10M<n<100M
---

# Recognize Anything 14M Dataset Card

## Dataset details

**Dataset type:**
These annotation files come from the Recognize Anything Model (RAM). RAM propose an automatic data engine to generate substantial image tags from image-text pairs.

**Dataset date:**
Recognize Anything 14M Dataset was collected in April 2023, by an automatic data engine proposed by RAM.

**Paper or resources for more information:**
https://github.com/xinyu1205/recognize-anything

**Where to send questions or comments about the model:**
https://github.com/xinyu1205/recognize-anything/issues

## Intended use
**Primary intended uses:**
The primary use of Recognize Anything is research on fundamental image recognition models.

**Primary intended users:**
The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.