---
license: mit
task_categories:
- image-classification
tags:
- scene-classification
- computer-vision
size_categories:
- 1K<n<10K
---

# Location Classification Dataset


## Dataset Summary
This dataset contains images extracted from videos for scene classification into 4 categories:
- Cafe
- Gym
- Library
- Outdoor

## Purpose
The dataset was created as part of a course project to perform location classification from an input image of the user's surroundings.
The dataset represents real-world indoor and outdoor environments with varying lighting conditions, angles, and compositions.

## Composition
- **Instances:** 503 total images  
- **Features:**  
- Label: One of the four labels mentioned prior 

## Collection
Original samples were collected manually for the assignment. Labels were assigned to each sample as categorical classes.
The images were collected from short 10-15 second videos at each location, and every 10th frame was extracted as an entry to prevent very similar frames

Format: JPEG
Images were preprocessed with resize to 224x224 and converted to tensors.

## Labels
- `Rating`: categorical class variable (multi-class).  
- Distribution approximately balanced via augmentation.


## Intended Uses
- Location classification from an input image
- 
## Limitations
- Locations were restricted to Carnegie Mellon University campus, so variability in its ability to recognize different versions of the environment may be hindered

## Ethical Considerations
- No personal or sensitive content included, but all images were collected from public areas.  

## License
CC-BY-4.0

## AI Usage
- AI tools (ChatGPT, Hugging Face transformers) assisted in drafting this card and extracting data