This dataset contains images of cars categorized into two classes: exterior and interior. It is designed for training and evaluating binary image classification models that can distinguish between the outside and inside views of a car. Images were collected via web scraping from publicly available sources.

Dataset Structure

Data Splits

Split	Number of Images
Train	710
Validation	178
Total	888

Data Fields

image — A PIL Image in RGB format
label — An integer class label:
- 0 → exterior
- 1 → interior

Class Distribution

Class	Label	Description
exterior	0	Images showing the outside body of a car
interior	1	Images showing the inside cabin/cockpit of a car

Dataset Creation

Source Data

Images were collected by web scraping from publicly available online sources such as car listing websites, automotive blogs, and image search engines.

Collection Process

Images were scraped using automated tools and then manually reviewed to ensure relevance and quality. Duplicate or low-quality images were removed during preprocessing.

Who Created This Dataset

This dataset was created for the purpose of training a car view classification model.

Uses

Direct Use

Binary image classification (exterior vs. interior)
Transfer learning and fine-tuning of image classification models
Automotive computer vision research and prototyping

Out-of-Scope Use

This dataset is not suitable for object detection or image segmentation tasks
Should not be used to make safety-critical decisions without further validation
Not intended for identifying specific car models, makes, or brands

Dataset Statistics

Total images: < 1,000
Number of classes: 2 (exterior, interior)
Image format: JPG / PNG
Image size: Varies (resize recommended before training)

Annotations

Labels are derived from the folder structure used during collection. Each image was manually verified to confirm it belongs to the correct class.

Bias, Limitations & Risks

Images were collected via web scraping and may reflect biases present in online automotive content (e.g. overrepresentation of certain car brands, colors, or styles)
The dataset is relatively small (~ 1,000 images), which may limit model generalization
Scraping sources may include images with varying lighting conditions, angles, and quality levels
Users should validate model performance on their own domain-specific data before deployment

Sample Images

Exterior	Interior

Downloads last month: 4