# Dataset

## Categories
- [x] - Gaming
- [x] - Video/Streaming
- [ ] - Texting
- [ ] - Online Course/Learning
- [ ] - Reading News/Articles (?)
- [ ] - Text Editing
- [ ] - Social Media
- [ ] - Idle

## Methods
- Existing Dataset
- Google Image Search
- Screenshot generator

## Kaggle Dataset 1
[13,000 screenshots manually categorized](https://www.kaggle.com/datasets/datasnaek/lightshot)

### Categories
* Animations
* Games
* Objects
* People
* Text
* Animated game character

### Result

85% accuracy for categorisation with the rest of the image set
- test set: 1,000,000 images
- training set: 5,609 images

## Kaggle Dataset 2
[20,000 webpage screenshots](https://www.kaggle.com/datasets/aydosphd/webscreenshots)

### Categories
* Machinery
* Sports
* Music
* People
5000 each

### Result
Not mentioned

## Hugging Face Gameplay Dataset
[6,020 gameplay screenshots](https://huggingface.co/datasets/asgaardlab/GameplayCaptions)

### Categories
* Gaming
6,020 images

### Result
Used for catogrizing different games, but we can use it as the gaming dataset

## Kaggle Face Gameplay
[10,000 gameplay screenshots](https://www.kaggle.com/datasets/aditmagotra/gameplay-images)

### Categories
* Gaming
10,000 images

### Result
Used for catogrizing different games, but we can use it as the gaming dataset

## Youtube Dataset
[YouTube8M Dataset](https://research.google.com/youtube8m/)

### Categories
* Videos

### Hugging Face Dataset
[Screenspot](https://huggingface.co/datasets/rootsautomation/ScreenSpot)

### Categories
- Not sure, probably files or webpage?