Youtube link - https://youtu.be/DCcP2pdPuG4

Rock-Paper-Scissors

Table of Contents

• Requirements
• Installation
• Usage
  • Classifying a Single Image
  • Real-time Webcam Classification
• Files Overview
• License

---

Requirements

Libraries installed:

• numpy
• opencv-python
• tensorflow
• matplotlib

You can install the dependencies using:

pip install numpy opencv-python tensorflow matplotlib

Installation

• Model file: keras_model.h5
• Label file: labels.txt

3. Update the paths in the code for both the model and label files as per your system configuration.

MODEL_PATH = r"C:\path\to\model\keras_model.h5"
LABELS_PATH = r"C:\path\to\model\labels.txt"

4. If you plan to run real-time classification, ensure you have a working webcam.

Usage

Classifying a Single Image

You can classify a single image by running rock-paper-scissor.py

python classify_image.py <image_path>

For example:

python classify_image.py C:\path\to\your\image.jpg

This script will:

• Load the specified image.
• Preprocess it (resize to 224x224 and normalize pixel values).
• Use the pre-trained model to predict the class (Rock, Paper, Scissors).
• Display the image with the predicted class and confidence score.

Real-time Webcam Classification

For real-time hand gesture classification using your webcam, run:

python rock-paper-scissor-live.py

This script will:

• Open the webcam and capture frames in real-time.
• Preprocess each frame (resize to 224x224 and normalize).
• Use the pre-trained model to predict the gesture.
• Display the webcam feed with the predicted class overlaid.
• Save the processed video to an output file (output.avi).

Video Saving

The webcam-based classifier also saves the video stream to an output file (output.avi) while displaying the classification results in real-time. You can modify the output file path in the code:

output_file = r"C:\path\to\save\output.avi"
Youtube link - https://youtu.be/DCcP2pdPuG4

Files Overview

• rock-paper-scissor.py: This script loads a single image from the specified path, processes it, and predicts the class (Rock, Paper, Scissors) using a pre-trained model. The prediction and confidence score are displayed along with the image.

teachable.ipynb: This ipynb loads a single image from the specified path, processes it, and predicts the class (Rock, Paper, Scissors) using a pre-trained model. The prediction and confidence score are displayed after the code cell is run.

• rock-paper-scissor-live.py: This script captures video from the webcam in real-time, processes each frame, and predicts the class (Rock, Paper, Scissors). The predicted class is displayed on the video feed, and the video is saved to an output file.

• labels.txt: This file contains the class names (e.g., Rock, Paper, Scissors) that correspond to the model's output classes.

• keras_model.h5: Pre-trained TensorFlow Keras model for classifying hand gestures.

Sample Code Structure

.
├── rock-paper-scissor.py          # Script to classify a single image
├── rock-paper-scissor-live.py    # Script to classify gestures from webcam feed
├── teachable.ipynb            # ipynb image for testing implementation of rock-paper-scissor code 
├── model
│   ├── keras_model.h5         # Pre-trained model (download separately)
│   └── labels.txt             # Class names for the model (download separately)
├── output.avi                 # Saved video file from webcam classification
└── README.md                  # Project documentation

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Example Command

To classify an image:

python rock-paper-scissor.py C:\path\to\image.jpg

To classify real-time video feed:

python rock-paper-scissor-live.py

---