International Journal of Advances in Engineering and Management (IJAEM)

Volume 5, Issue 4 April 2023, pp: 1894-1896 [Link] ISSN: 2395-5252

Hand Gesture Volume Control Using

Machine Learning
[Link] Emelda V, Shivani, Sanskruti S
Assistant Professor, department of Computer Science Engineering, MVJ College of Engineering, Bangalore,
Karnataka, India.
[Link],Computer science engineering, MVJ College of Engineering, Bangalore, Karnataka,India.

----------------------------------------------------------------------------------------------------------------------------- ---------
Date of Submission: 20-04-2023 Date of Acceptance: 30-04-2023
----------------------------------------------------------------------------------------------------------------------------- ----------
ABSTRACT keyboard, mouse, joystick and touch screen, but
A systems that can recognize a hand motion in real these devices does not provide easier way to
time video is hand gesture recognition. Hand communicate. In this, the system which is proposed
gestures are categorized according to their subject will consists of desktop and laptop interface, hand
matter. The design of hand gesture recognition is gesture can be used by the users need to wear data
one of the More difficult jobs, as it combines two gloves also can use the web camera or separate
significant issues. The detection of the hand is the cameras for recording the hand gestures.
first step and creating a sign that can only be The first and most important step toward
utilized by one hand at a time. It can be used in a any hand gesture recognition system is to
variety of settings, including human-computer implement hand tracking system. Some Sensor
interaction and sign language. The basic concept of devices are generally used in DataGlove based
hand segmentation and the hand detection system, methods for digitizing hand and finger motions into
which use the Haar-cascade classifier, may be used multi parametric data. Other sensors used in this
to construct hand gesture recognition using Python system will collect hand configuration and hand
and OpenCV. We discuss a way for hand gesture movements.
identification based on shape-based features The vision Based method requires a web
detection. The configuration comprises a single camera, so that one can realize natural interaction
camera that captures the user's gesture and feeds it between humans and computer without using any
into the system. A fundamental goal of gesture other devices. The challenging part in these
recognition is to develop a system that can systems is background images or videos which is
recognize specific human gestures and utilize them recorded or captured during taking the inputs i.e.
to send information for device control. With real- hand gesture by the user, also sometime lightning
time gesture recognition, a user can operate a effect the quality of the input taken which creates
computer by making a specific gesture in front of a the problem in recognizing the gestures. Process to
computer's camera. With the help of the OpenCV find a connected region within the image with
module, we will create a hand gesture. Without the some of the property such as color ,intensity and a
use of a keyboard or mouse, the system can be relationship between pixels i.e. pattern is termed as
controlled via hand gestures. segmentation. And have used some important
packages which have OpenCv-python, tensorflow,
INDEXTERMS numpy, mediapipe, imutils, scipy,numpy.
 Camera input
PROCESS OF HAND GESTURE VOLUME
 Hand tracking
CONTROL
 Feature extraction • Download and import these libraries – Mediapipe,
 Gesture recognition OpenCV, Time, Math, Numpy, ctypes, comtypes
and pycaw
I. INTRODUTION • Turn on the webcamera
Hand gestures is the powerful • Read all the frames one-by-one
communication medium for Human Computer • Detect all the hands (1 or more, if any) in the
Interaction (HCI).Several input devices are frame
available for interaction with computer, such as • Get all the hand-landmarks for all the detected

DOI: 10.35629/5252-050418941896 |Impact Factorvalue 6.18| ISO 9001: 2008 Certified Journal Page 1894
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 5, Issue 4 April 2023, pp: 1894-1896 [Link] ISSN: 2395-5252

hands classification, by which a data point's classification

• If the number of detected hands is more than one, is determined by how its neighbor is classified.
then consider only one hand for the project. Euclidean distance is used to find the nearest
• On this hand, identify the index finger and the neighbor in KNN.
thumb. Image processing is a method to perform
• Find the distance between the tips of the index some operations on an image, in order to get an
finger and the thumb enhanced image and or to extract some useful
• Access the volume-controlling ability of your information from it. If we talk about the basic
system by using the ctypes library which uses definition of image processing then “Image
functions cast and SetMasterVolumeLevel to processing is the analysis and manipulation of a
control the volume of any system. digitized image, especially in order to improve its
• Bring the tips of both the index finger and the quality”. Image processing is basically signal
thumb together and then away from each-other. processing in which input is an image and output is
• Using the Interpolation function offered by image or characteristics according to requirement
Numpy, we can convert the changing distance associated with that image. Image processing
between the index finger and the thumb (got by the basically includes the following three steps:
movement of the two fingers), into the volume * Importing the image
level of our system. Farther the tips of the two * Analysing and manipulating the image
fingers from each-other, higher is the volume level, * Output in which result can be altered image or
which means, nearer the tips of the two fingers, report that is based on image analysis.
lower is the level of the volume in our device. Mediapipe is an open source cross-
• Display the video captured by the web-camera, by platform framework provided by Google to build a
displaying all the frames one after the other, pipeline for processing perceptual data from
thereby, displaying our hand and the action different modalities such as video and audio. The
performed by the index finger and the thumb. solutions used in MediaPipe include multiple items
such as posture estimation and face recognition. In
PHASES IN DETECTING HAND GESTURE this paper, we will use MediaPipe Hands for hand
In vision community hand gesture is an tracking.
active area of research, for the purpose of sign
language recognition and human computer Advantages
interaction. In this we have used some algorithms 1. Easy to use.
and some modules to detect the gestures of the 2. Earlydetection.
person and these gestures are taken as the input in
the system . Here, several modules are used like
3. Hassle free.
opencv-python, mediapipe, numpyetc for the 4. Time-efficient.
purpose of tracking the gestures. After capturing 5. More interactive.
the input from the user the image is used in the Disadvantages
hand tracking system to check the dimensions and 1. Technicalchallenges.
shape of the gesture which is received in the 2. Requires a decent camera.
system. Hand tracking module plays a important 3. They are neither self revealing nor self
role in identifying the input recorded in the system, explanatory.
after that classification and segmentation process is
4. May be confused by two palms.
used to classify the gestures in the system .Machine
learning and deep learning is also used to identify
the training data from the system and identify it
III. FUTURE SCOPE
The model successfully controls the
according to the requirement of the system .After
volume of the system by hand gesture, that means
this the gestures are identified from the trained data
the volume of the device is controlled without
and on the basis of that data the gestures rae
opening the volume-settings or the volume icon in
recognized and is used for processing of the the
the system. The accuracy of the model is
system to implement the functions like increase and
100%.This model successfully controls the volume
decrease in volume.
of a device without a mouse, and without having to
open the volume icon or the volume settings. The
II. METHODOLOGY model uses the index finger and the thumb of our
For gesture recognition, the K-Nearest
hand, and a camera. The model uses hand-
Neighbor (KNN) algorithm is a supervised
landmarks defined by Google‟s Mediapipe library,
machine-learning algorithm. KNN is used for

DOI: 10.35629/5252-050418941896 |Impact Factorvalue 6.18| ISO 9001: 2008 Certified Journal Page 1895
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 5, Issue 4 April 2023, pp: 1894-1896 [Link] ISSN: 2395-5252

and the ctypes library, for seamless execution. and gesture recognition.1995, 12: 296-
301.
IV. RESULT AND CONCLUSION [7]. G. R. S. Murthy, R. S. Jadon. (2009). “A
This project is presenting a program that Review of Vision Based Hand Gestures
allows the user to perform hand gesture for Recognition,” International Journal of
convenient and easier way to control the software Information Technology and Knowledge
.A gesture based volume controller doesn‟t require Management, vol. 2(2), pp. 405410.
some specific type of markers and these can be [8]. Mokhtar M. Hasan, Pramoud K. Misra,
operated in our real life on simple Personal (2011). “Brightness Factor Matching For
Computers with a very low cost cameras as this not Gesture Recognition System Using Scaled
requires very high definition cameras to detect or Normalization”, International Journal of
record the hand gestures. Specifically, system Computer Science & Information
tracks the tip positions of the counters and index Technology (IJCSIT), Vol. 3(2)
finger of each [Link] main motive of this type of [9]. Bartlett, J.F.: Rock „n‟ Scroll Is Here to
system is basically to automate the things in our Stay. IEEE Computer Graphics and
system in order to make the things become easier to Applications (2000)Google Scholar
control. So in order to make it realiable we have [10]. Analog Devices, ADXL202E: Low-
used this system to make the system easier to Cost?2g Dual-Axis Accelerometer With
control with the help of these [Link] this Duty Cycle Output Data Sheet (Rev. A,
paper we mainly focused on detecting the hand 10/00).
landmarks and video cam. After which we can
access the audio functions inside the system and
based on the distance between the thumb and the
index finger there will be alteration in the volume.

REFERENCES
[1]. Akira Utsumi, TsutoniuMiyasato, Fumio
KishinoandRyoheiNakatsu, “Real-time
Hand Gesture RecognitionSystem,” Proc.
of ACCV ‟95, vol. 11, pp. 249-253,
Singapore,1995
[2]. Attila Licsár, TamásSzirányi University of
Veszprém, “Dynamic Training of Hand
Gesture Recognition System” Department
of Image Processing andNeurocomputing,
H8200 Veszpré,23-26 Aug. 2004
[3]. L. Bretzner and T. Lindeberg, “Relative
orientation from extended sequences of
sparse point and line correspondences
using the affine trifocal tensor,” in Proc.
5th Eur. Conf. Computer Vision, Berlin,
Germany, June 1998, vol. 1406, Lecture
Notes in Computer Science, pp.141–157,
Springer Verlag
[4]. Intel Corp, “OpenCV Wiki,” OpenCV
Library [Online], Available:
[Link] /
[5]. Z. Zhang, Y. Wu, Y. Shan, S. Shafer.
Visual panel: Virtual mouse keyboard and
3d controller with an ordinary piece of
paper. In Proceedings of Perceptual User
Interfaces, 2001
[6]. W. T. Freeman and M. Roth, Orientation
histograms for hand gesture recognition.
International workshop on automatic face

DOI: 10.35629/5252-050418941896 |Impact Factorvalue 6.18| ISO 9001: 2008 Certified Journal Page 1896