theguywhosucks/basic500

---
license: agpl-3.0
task_categories:
- image-classification
tags:
- color
- bounding-box
pretty_name: basic
size_categories:
- 100M<n<1B
---

<p align="center">
<img src="banner.png" alt="basic500 banner" width="670"/>
</p>

<p align="center">
<a href="https://huggingface.co/datasets/theguywhosucks/basic500">
<img src="https://img.shields.io/badge/HuggingFace-FFD21E?style=for-the-badge&logo=huggingface&logoColor=black"/>
</a>
<a href="#">
<img src="https://img.shields.io/badge/Documentation-1E90FF?style=for-the-badge&logo=googlechrome&logoColor=white"/>
</a>
</p>

> **basic500** is a massive, high-quality dataset designed for training and benchmarking computer vision models in object detection, segmentation, and other vision tasks.
> *Created by **Mafulabs**.*

## 🎯 Key Features

* **Huge Scale:** 500 million color images
* **Diverse Objects:** Covers a wide range of real-world items
* **High-Quality Annotations:** Precise bounding boxes with minimal overlap
* **Optimized for ML:** Structured for blazing-fast access and framework integration

## 💾 Dataset Info

* **Images:** 500,000,000
* **Resolution:** High-quality, varied
* **Annotations:** Bounding boxes with class labels
* **Use Cases:** Object detection, segmentation, AI pretraining

## ⚙️ Getting Started

First, pull the dataset down from Hugging Face:

```bash
git clone https://huggingface.co/datasets/theguywhosucks/basic500
cd basic500

```

Then, load it up in your Python environment:

```python
from basic500 import load_dataset

# Load the dataset
dataset = load_dataset(path="path/to/basic500")

# Access images and bounding boxes
for img, bboxes in dataset:
display(img)
print(bboxes)

```

## 🧩 Tech Stack

| Technology | Purpose |
| --- | --- |
| 🐍 **Python** | Data processing and scripts |
| 🤗 **Hugging Face** | Dataset hosting and access |
| 🧮 **NumPy / OpenCV** | Image manipulation |

## 📜 Citation

If you use this dataset in your work, please cite it:

```bibtex
@misc{basic500_dataset,
title = {basic500: 500 Million Color Image Bounding Box Dataset},
author = {Mafulabs},
year = {2026},
howpublished = {\url{[https://huggingface.co/datasets/theguywhosucks/basic500](https://huggingface.co/datasets/theguywhosucks/basic500)}}
}

```

## ⚖️ License

Distributed under the **AGPL-3.0** License. Read the `LICENSE` file for more info.

---

*Designed for researchers, developers, and hobbyists seeking large-scale, high-quality image datasets for AI vision tasks.*