RPS_project

“Rock paper scissors” is a simple and interesting game. Many of us used to play it in a school to resolve disputes or just to spend some time. But how to play it with computer using gesture?

The rules are very simple and probably you remember them from your childhood:

• If you choose the same action, it is a tie.
• If you choose Rock, you will win against Scissors but lose against Paper.
• If you choose Scissors, you will win against Paper but lose against Rock.
• If you choose Paper, you will win against Rock but lose against Scissors.

If you are ready to start, just press key "s" to start, then there will be a five seconds countdown. The game ends when somebody gets 3 wins. If you want to play again, just press key “c” after a game is finished.

Milestone 1

In this milestone, I ues the Teachable Machine to create machine learning models for Rock, Paper, Scissors. Teachable Machine is a web-based tool which can be used to create machine learning models for images,sounds and poses.

Just simpily click the Webcam icon and record gestures of Rock, Paper, Scissors. The "Nothing" class represents the lack of option in the image. After training, download the model from the "Tensorflow" tab in Teachable-Machine.

Milestone 2

In this milestone, I install the dependencies for the game "Rock, Paper and Scissors". To separate dependencies of different projects, I creat isolating environment for this project and then activate it. The commands on terminal are as follow:

Create new environment with python:

conda create --name <env_name> python=3.8

Activate environment

conda activate <env_name>

Install pip,opencv,tensorflow and ipykernel

conda install pip
pip install opencv-python
pip install tensorflow
pip install ipykernel

Note: Most of the time there is no difference in installing them with conda or pip. The pip catalog is more complete, while the conda dependency resolver is more robust. Inside a conda virtual environment, you have access to both conda and pip to install libraries.

Milestone 3

In this milestone, I write the game logic of Rock-Paper-Scissors. However, in this stage, the game just request input from user's keyboard, rather than the camera. There are two main tasks as follows:

1. Store the user's and computer's choices

Define a list ["Rock","Paper","Scissors"]. Call the function random.choice(list) to choose a random action for computer. Then, call input() to request a input from user. To prevent spelling mistake, just ask an integer and use the index to find the corresponding action in list.

2. Figure out who won

Using if-else statements to choose a winner based on the classic rules of Rock-Paper-Scissors. If the action of user is the same as the computer's action, it is a tie and nothing will happen. If the action of user is one of Rock,Paper or Scissors, the computer will have two different actions. print the winner under different situation according to the game rules.

Milestone 4

In this milestone, I combine the code that used the webcam with the function that ask user for an input. Replace the manual input for the output of the RPS model. In addtion, I add three code part to improve the interaction of the game application. These parts are as follows:

1. Count down

To give user enough time to prepare action, I create a countdown to get ready for the game. I set a countdown variable name TIMER, TIMER will decrease by one every second. When the TIMER = 0, the webcam will capture the image and pass this image to RPS model to predict the user's action. In function make_prediction() the computer will choose a random action so that I can judge if the computer win or the user win through the game rules.

2. Repeat until a player gets three victories

To implement this part, I set two global variables named usr_win_count and cp_win_count to calculate the victory times for user and computer. If it is a tie, these two variable will not change. If the user win, usr_win_count will increase by 1, otherwise cp_win_count increases by 1. After every round, judge if user or computer already won three times.(usr_win_count = 3 or cp_win_count = 3)

3. Press c to continue To print the message in the webcam display, I use cv2.putText() to add the text "Press c to continue". In order to react when pressing 'c',I call the cv2.waitkey() function. waitkey() function of Python OpenCV allows users to display a window for given milliseconds or until any key is pressed. So I use the following code to detect keypress.

if cv2.waitKey(1) & 0xFF == ord('c'):
    usr_win_count = 0
    cp_win_count = 0

waitKey() returns a 32 Bit integer value (might be dependent on the platform). The 0xFF in this scenario is representing binary 11111111 a 8 bit binary, since we only require 8 bits to represent a character. This is the reason why doing a AND operation between waitKey() to 0xFF.

Note: waitKey(0) will pause the screen because it will wait infinitely for keyPress on the keyboard and will not refresh the frame(cap.read()) when using the webcam. However, waitKey(1) will wait for keyPress for just 1 millisecond and it will continue to refresh and read frame from webcam using cap.read().