We present the findings and outcomes of our image classification model developed using Convolutional Neural Networks (CNN) for the Rock, Paper, Scissors game using streamlit app.
We have an interactive app where users can play the game against the computer by opening the webcam, making a hand gesture, and the model would predict the corresponding choice.
The model is integrated into an app to provide an interactive gaming experience.
It is important to note that our model's accuracy was influenced by the limitations imposed on uploading trained model size. We were not able to upload a more deep trained model because it exceeded 25Mb which is the Github limit. Consequently, we optimized our model to achieve the highest possible accuracy while ensuring the model size remained within the restrictions.
Within the other repo, we also provided a better model with higher accuracy, but we were not able to implement the outcome of that model to use it in the app.
Overall, our project successfully showcased the capabilities of image classification using CNNs, but not accurate enough. Our image classification model demonstrated promising performance in predicting the hand gestures of rock, paper, and scissors when provided with uploaded images. While we faced limitations in implementing real-time webcam-based interactions, the app still offered an engaging user experience.
This is the link of the app:
https://rockpapergame.streamlit.app/

And this is a screen shot of app working properly.
https://raw.githubusercontent.com/Kawians/StreamlitProject/main/Rock-paper-scissors.jpg