Scribd Logo
Upload

Welcome to Scribd!

  • Hand Gesture Recognition Using CNN

    Uploaded by

    musaveer khan
    33 views9 pages
    This document describes a hand gesture recognition system using a convolutional neural network (CNN). The objectives are to implement hand detection, feature extraction, gesture detection, and an application. The methodology uses input images, pre-processing, a CNN configuration with convolution, pooling and fully connected layers, and a two-phase training process. Pending work includes expanding the dataset, a literature review, identifying key features, understanding existing approaches, and developing a web application.

    Original Description:

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    This document describes a hand gesture recognition system using a convolutional neural network (CNN). The objectives are to implement hand detection, feature extraction, gesture detection, and an application. The methodology uses input images, pre-processing, a CNN configuration with convolution, pooling and fully connected layers, and a two-phase training process. Pending work includes expanding the dataset, a literature review, identifying key features, understanding existing approaches, and developing a web application.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Flag for inappropriate content
    33 views9 pages

    Hand Gesture Recognition Using CNN

    Uploaded by

    musaveer khan
    This document describes a hand gesture recognition system using a convolutional neural network (CNN). The objectives are to implement hand detection, feature extraction, gesture detection, and an application. The methodology uses input images, pre-processing, a CNN configuration with convolution, pooling and fully connected layers, and a two-phase training process. Pending work includes expanding the dataset, a literature review, identifying key features, understanding existing approaches, and developing a web application.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Flag for inappropriate content
    of 9

    Hand Gesture

    Recognition Using
    Batch No – B24
    CNN Under the guidance of :
    MUSAVEER KHAN 4MH18CS064 Prof. Anand Kumar H N
    RUCHITHA M 4MH18CS091 Dept. of CSE.
    SATHWIKNIDHI Y T 4MH18CS099
    YADUNANDAN A R 4MH19CS412
    Introduction
    • Bobick and Wilson have given a definition of gesture.

    • Gestures may be classified as dynamic or static. A dynamic gesture intends to vary over a period of time and

    static gesture tends to remain almost unchanged over time.

    • The key problem is how to make a computer able to understand the hand gestures.

    • The process is the integration of two tasks: feature extraction and classification.

    • Huge features are mandatory for classification and recognition. Conventional models for pattern recognition

    cannot process natural data in raw form.


    Objectives
    The system objectives are created throughout the project. They resemble the vital outputs
    which are needed to successfully create the system.

     Successfully implement a hand detection technique to identify the symbols shown in hand.

     Successfully implement a method to extract useful features.

     Successfully implement a method to detect gestures.

     Create an application that shows the capabilities of the system.


    Methodology
    This section provides the description of the dataset and CNN configuration that are using. The
    flowchart of methodology is shown on Figure . The approach is the combination of data
    collection, pre-processing, configuring the CNN and building the model.
    1. Input Data and Training Data
    • Images needed to train and validate the model are collected using a webcam.
    • The recognition process will be less complex and more efficient if the background is less complex
    and the contrast is high on the hand.
    • So, it is assumed that the background of the images was less complex and uniform.

    2. Pre-Processing
    • A minimal pre-processing was applied over the dataset to reduce the computational complication and
    achieve better efficiency.
    3. Dataset
    • We selected 10 static gestures (Index, Peace,
    Three, Palm Opened, Palm Closed, OK,
    Thumbs, Fist, Swing, Smile) to recognize.

    4. CNN Configuration
    • The CNN has been considered in this research to recognize hand gesture is composed of two
    convolution layers, two max pooling layers, two fully connected layers and output layer.
    • There are three dropout performance in the network to prevent over-fitting.
    5. System Implementation
    • To implement the system, python was used as the programming language
    and a python IDE Spyder was used to write and run code.
    • The training process on dataset is composed of two phases.
     Training with Base Dataset.
     Training with Expanded Dataset.
    Pending Work
    • Dataset Creation.
    • Detailed literature survey on existing approaches.
    • Identifying features which are most important to recognize
    the symbol precisely.
    • To understand existing system strongly so that we can
    bring up the best in proposed system.
    • Development of web based application.
    Thank You

    Q&A

    You might also like