Scribd
Upload
6 views25 pages

10th Batch Final Review

The document presents a research project on hand sign recognition using deep learning techniques, specifically employing MediaPipe to achieve accurate gesture recognition for communication. It outlines the methodology, dataset, and results, highlighting high accuracy for most gestures tested. The study aims to enhance interaction between humans and computing devices through gesture recognition, particularly for the hearing and speaking impaired.

Uploaded by

sowmiyasreem.24mca
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
6 views25 pages

10th Batch Final Review

The document presents a research project on hand sign recognition using deep learning techniques, specifically employing MediaPipe to achieve accurate gesture recognition for communication. It outlines the methodology, dataset, and results, highlighting high accuracy for most gestures tested. The study aims to enhance interaction between humans and computing devices through gesture recognition, particularly for the hearing and speaking impaired.

Uploaded by

sowmiyasreem.24mca
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 25

Paper ID:ICAIS-298

Title: HAND SIGN RECOGNITION


USING DEEP LEARNING
TECHNIQUES

GUIDED BY:
Ms.M.N.KAVITHA M.E.,

Team Members

Kaviyarasu S (21BIR022)
Ragunath N (21BIR036)
Rahul P (21BIR037)
Table of contents:
 Abstract

 Introduction
 Literature Review

 Problem Definition

 Dataset Description
 Workflow of the proposed model

 Proposed Methodology
 Conclusion

 References
Abstract:
 In recent years Deep Learning techniques have demonstrated
remarkable performance in various tasks, including object recognition
and segmentation.
 Our research project focuses on the application of Deep Learning
methods to recognize the hand sign gesture’s.
 The objective of this study is to design and implement a accurate hand
sign gesture recognition system using MediaPipe.
 We are using 10 hand gestures for rapid message communications.
Introduction:
 Early research on hand shape and gesture recognition took

place between the 1960s and the 1990s.


 Gestures are a powerful means of communication among

humans. In fact, gesturing is so deeply rooted in our


communication.
 A natural interaction between humans and computing devices

can be achieved by using hand gestures for communication


between them.
 We are using a Predefined framework Mediapipe to build our

model.
Literature Survey:
S.No Name of the Author Year of Concepts/ Accuracy
Paper Name publication Algorithm
used

1. Hand Gesture Indriani 2021 Mediapipe 95%


Recognition for Moh.Harris Framewor
User Guide Ali k
Application Suryaperda
Using MediaPipe na Agoes

2. Hand Gesture HIRA ANSAR, 2023 Convex 93.2%


Recognition for NAIF AL
MUDAWI,
Hull
Characters Landmarks
SAUD S.
Understanding ALOTAIBI ,
Using Convex ABDULWAHA
Hull Landmarks B ALAZEB ,
and Geometric BAYAN
Features IBRAHIMM
ALABDULLAH
,
MOHAMMED
ALONAZI,
AND
JEONGMIN
S.No Name of the Author Year of Concept Accuracy
Paper Name publicatio s/
n Algorith
m used

3. Deep learning Oyebade 2017 Convolutio 91.33


in vision-based K. nal neural
static hand Oyedotun, network
gesture Adnan and
recognition Khashman Stacked
denoising
autoencod
er
4. Hand gesture Gongfa Li, 2019 convolutio 98.52
recognition · Heng n neural
based on Tang · network
convolution Ying Sun, ·
neural Jianyi
network Kong,·
Guozhang
Jiang,· Du
Jiang · Bo
Tao, ·
Shuang
Xu, ·
S.No Name of Author Year of Concepts/ Accurac
the Paper Name publicati Algorith y
on m used
5. Vision Based Pragati 2009 3d and -
Hand Garg, appearance
Gesture Naveen based
Recognition Aggarwal approaches
and
Sanjeev
Sofat
Problem Statement:

Create a real-time hand gesture recognition


system for hearing and speaking impaired
with high accuracy, low latency, and
robustness in varying environmental
conditions.
Dataset Description:
As we are using the vector points that are recognized from the hand solution
of the mediapipe we are checking for a condition for the recognition of the
gesture recognition.
“OK” “THAT’S IT”
“I’LL TRY” “MOVE ON”
“YES I’LL” “NOT OK”
“IT’S WRONG” “ONCE AGAIN”
“BREAK” “I CAN’T”
Workflow of the Proposed Model:
Proposed Methodology:
Methods used:
1. OpenCV
2. MediaPipe

OpenCV:
 OpenCV stands for Open Source Computer Vision it is a powerful open-source
software library used for image and video processing tasks. Initially it was developed
by Intel, it offers a variety of functionalities, including computer vision algorithms,
machine learning integration, 3D vision capabilities, and more.
 Its cross-platform interoperability enables developers to work easily across Windows, Linux,

macOS, Android, and iOS, and the availability of bindings for several programming
languages, including Python, makes it easy to integrate into a wide range of applications.

Uses:
 Image and Video Processing: OpenCV provides tools to

perform operations on images and videos, including


filtering, transformation, and resizing, making it
essential for image and video processing tasks.
 Object Detection and Recognition: OpenCV is commonly

used for detecting and recognizing objects in images


and videos, enabling applications like face detection,
object tracking, and augmented reality.
MediaPipe:

 MediaPipe is an open-source, cross-platform framework


developed by Google for building real-time, cross-platform AI
and computer vision applications. For tasks like object
detection, face detection, hand tracking, position estimation,
and more, it offers an extensive collection of tools and solutions.
 With MediaPipe, developers can easily construct apps that

process and analyze visual data in real-time and efficiently—


whether it comes from videos, images, or live camera feeds.
Workflow of the Mediapipe :
Uses:

 MediaPipe Solutions provides a lot of libraries and tools to

quickly apply artificial intelligence (AI) and machine learning


(ML) techniques in our applications.
 We can use these solutions by customizing them to our needs,

and can be used across multiple development platforms.


Result:
HAND SIGN ACCURACY
CLASSIFICATION

OK 99%
THAT’S IT 99%
I’LL TRY 99%
MOVE ON 96%
YES I’LL 90%
NOT OK 99%
IT’S WRONG 99%
ONCE AGAIN 98%
BREAK 95%
I CAN’T 87%
Conclusion:
 Unlike traditional input / interaction with computer machines we have

used a new interaction concept through gestures.


 The proposed method using MediaPipe as one framework based on

machine learning plays an effective role for hand gesture recognition.


 Among the ten gestures considered , accurate results were achieved

for almost eight gestures.


 We would like to extend our system further with different algorithms

and additional gestures and experiment with both static and dynamic
hand gesture recognition systems.
References:
[1] Indriani1 Moh.Harris Ali Suryaperdana Agoes,” Applying Hand Gesture
Recognition for User Guide Application Using MediaPipe”
[2] HIRA ANSAR , NAIF AL MUDAWI , SAUD S. ALOTAIBI , ABDULWAHAB
ALAZEB , BAYAN IBRAHIMM ALABDULLAH , MOHAMMED ALONAZI , AND
JEONGMIN PARK,” Hand Gesture Recognition for Characters Understanding Using
Convex Hull Landmarks and Geometric Features “
[3] Oyebade K. Oyedotun • Adnan Khashman,” Deep learning in vision-based static
hand gesture recognition “
[4] Gongfa Li · Heng Tang· Ying Sun· Jianyi Kong · Guozhang Jiang · Du Jiang2 · Bo
Tao · Shuang Xu · Honghai Liu,” Hand gesture recognition based on convolution
neural network”
[5] Pragati Garg, Naveen Aggarwal and Sanjeev Sofat ,” Vision Based Hand Gesture
Recognition”
THANK YOU

You might also like