This project was created for #SchoolofAIVancouver Image Classification Code Challenge 2. This image classification code tries to classify video games into 3 classes:
- First Person Shooter(FPS)
- Real Time Strategy(RTS)
- Sports(SPT)
You can clone the repo to retrain and test the model. However please not that there arent any 'train', 'validate' or 'test' data in this repo. It has been removed to avoid uploading all the images to git.
Look at the 'ytvid-to-jpg' script to help with the data scraping
The video_game_classify.ipynb
model training and classfiying screenshot
You will need the data which you can get using the video_game_classify.ipynb
-
Google Colaboratory- Free Jupyter notebook enviroment hosted in the cloud. Developed by Google.
-
Python3 - The programming language used
-
Keras - Keras: A high-level neural networks API running TensorFlow
-
ytvid-to-jpg - YTVID-TO-JPG: A custom python script that downloads Youtube video based on queries and converts it images.
There are No known bugs in the video_game_classify.ipynb
at this moment.
The trained model in this image classifier is not perfect! The data used in this project was scraped from the Internet(Youtube). The dataset has not gone through rigorous data cleaning or parameter tuning techniques and there maybe some false positives during the labelling test. Feel free to retrain the model with a more accurate dataset.
- Guru Shiva - Initial work
This project is licensed under the MIT License
- Siraj Raval for ML/AI inspiration and starting School of AI
- Akshi Chaudary- SchoolofAIVancouver
- Johannes Harmse- SchoolofAIVancouver
- Xinbin Haung- SchoolofAIVancouver
- Billie Thompson- For this amazing README.md template
- SchoolofAIVancouver
- The Internet