---
id: rf100vl
version: "1.1.2"
license: unclear
license_treatment: permissive
maintenance: active
---

# rf100vl — RF100-VL Dataset Interface

License: permissive · Maintenance: active · Downloads: 190.6K/mo

## What it is and what it does

RF100-VL is a Python interface to a large-scale object detection benchmark comprising 100 annotated datasets spanning seven domains: flora and fauna, sport, industry, document processing, laboratory imaging, aerial imagery, and miscellaneous real-world use cases. The package wraps the Roboflow platform to provide helper functions for retrieving dataset objects and downloading data in COCO JSON format.

The package is designed for researchers and practitioners benchmarking object detection models, including fully supervised, semi-supervised, few-shot, and vision-language models with localization capabilities. It includes subsets like RF100-VL-FSOD and RF20-VL for few-shot object detection evaluation, and supports the CVPR 2025 Few-Shot Object Detection challenge. Datasets are downloaded to local directories organized by project, making it straightforward to iterate on model development and evaluation.

Use it for:

- Benchmark vision-language models on out-of-distribution object detection tasks across multiple domains.
- Evaluate few-shot object detection models using RF20-VL-FSOD with 10 annotated examples per class.
- Download and organize multi-domain datasets for training and testing custom object detection systems.
- Participate in the CVPR 2025 Few-Shot Object Detection challenge with standardized evaluation protocols.
- Compare fully supervised and semi-supervised detection approaches on diverse real-world imagery.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides programmatic access to RF100-VL, a multi-domain object detection benchmark dataset of 100 datasets across seven domains, enabling download and evaluation of vision-language models and object detection systems.

Yes. The package offers low install friction, active maintenance, permissive licensing, and provides access to a substantial benchmark resource for object detection research. Install it if you are working on vision-language models, few-shot detection, or need a diverse multi-domain evaluation dataset—but note that you must obtain and configure a Roboflow API key beforehand.

## Install

pip install rf100vl
uv add rf100vl
poetry add rf100vl

## Installing rf100vl

Before you install: Low install friction with a single runtime dependency (roboflow). Package is actively maintained with recent releases.

License in practice: Licensed under permissive terms (MIT), allowing commercial and private use without significant restrictions.

Quickstart:

pip install rf100vl

from rf100vl import download_rf100vl

download_rf100vl(path="./rf100-vl/")

Requires a free Roboflow Universe API key set in the ROBOFLOW_API_KEY environment variable to download datasets.

Verify before relying:

- Whether the package supports incremental or resumable downloads for large dataset transfers.
- Performance characteristics when downloading multiple datasets or subsets in parallel.
- Memory requirements for loading and processing full RF100-VL or RF20-VL datasets.

## Package facts

- License: not declared (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 190.6K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

object detection benchmark dataset, vision language model evaluation, multi-domain detection data, roboflow dataset download, few-shot object detection benchmark, computer-vision, benchmark-dataset, object-detection

[View on SkillFed](https://skillfed.io/packages/rf100vl) · [View on PyPI](https://pypi.org/project/rf100vl/)