SCT_ML_4 - Hand Gesture Recognition using CNN

Project Overview

This project implements a Hand Gesture Recognition System using a Convolutional Neural Network (CNN). The model is trained on the LeapGestRecog Dataset and can classify different hand gestures from image data. The goal of this project is to enable gesture-based human-computer interaction.

Features

• Hand Gesture Classification using CNN
• Image Preprocessing with OpenCV
• Gesture Distribution Visualization
• Model Accuracy Visualization
• Model Loss Visualization
• Gesture Prediction with Confidence Score
• Trained Model Generation

Dataset

Dataset: LeapGestRecog Dataset

The dataset contains the following hand gestures:

• Palm
• L
• Fist
• Fist Moved
• Thumb
• Index
• OK
• Palm Moved
• C
• Down

Technologies Used

• Python
• TensorFlow
• Keras
• OpenCV
• NumPy
• Matplotlib
• Seaborn
• Scikit-learn

Project Structure

SCT_ML_4/
│
├── hand_gesture_recognition.py
├── requirements.txt
├── README.md
├── gesture_distribution.png
├── accuracy_graph.png
├── loss_graph.png
├── gesture_prediction.png
└── hand_gesture_model.h5

Installation

Install the required libraries:

pip install -r requirements.txt

Run the Project

python hand_gesture_recognition.py

Output

The project generates:

• Gesture Distribution Graph
• Accuracy Graph
• Loss Graph
• Gesture Prediction Result
• Trained CNN Model

Results

Gesture Distribution

Model Accuracy

Model Loss

Gesture Prediction

Learning Outcomes

Through this project, I gained practical experience in:

• Deep Learning
• Convolutional Neural Networks (CNN)
• Computer Vision
• Image Classification
• Data Preprocessing
• Model Evaluation

Acknowledgement

This project was completed as part of the Machine Learning Internship Program at SkillCraft Technology. It provided valuable hands-on experience in deep learning and computer vision applications.