README.md

---
license: other
task_categories:
- image-classification
language:
- en
pretty_name: Intel Image Classification
size_categories:
- 10K<n<100K
---

# Intel Image Classification

The **Intel Image Classification** dataset contains images of natural scenes categorized into six classes:

- Buildings

- Forest

- Glacier

- Mountain

- Sea

- Street

---

## 📆 Content

- The dataset contains **~25,000 images** of size **150x150 pixels**.

- Images are evenly distributed across **6 categories**:
  
  ```
  {'buildings' -> 0,
   'forest'    -> 1,
   'glacier'   -> 2,
   'mountain'  -> 3,
   'sea'       -> 4,
   'street'    -> 5 }
  ```

- It is divided into three parts:
  
  - **Training set**: ~14,000 images
  
  - **Test set**: ~3,000 images
  
  - **Prediction set**: ~7,000 images

The train, test, and prediction images are stored in separate folders.

---

## 🧪 Structure

```
data/
├── seg_train/
│   ├── buildings/
│   ├── forest/
│   ├── glacier/
│   ├── mountain/
│   ├── sea/
│   └── street/
├── seg_test/
│   └── ...
└── seg_pred/
    └── ...
```

---

## 🔗 Source & Acknowledgements

- Originally published by **Intel** as part of a challenge on **Analytics Vidhya**:  
  [https://datahack.analyticsvidhya.com](https://datahack.analyticsvidhya.com/)

- Rehosted on Kaggle:  
  [Intel Image Classification | Kaggle](https://www.kaggle.com/datasets/puneet6060/intel-image-classification)

---

## 💻 Usage

You can load this dataset using Hugging Face's `datasets` library:

```python
from datasets import load_dataset

dataset = load_dataset("sfarrukhm/intel-image-classification")
```