VIRTUAL PAINT APPLICATION USING HAND GESTURES

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 3090
VIRTUAL PAINT APPLICATION USING HAND GESTURES
Niharika M1, Neha J2, Mamatha Rao3, Vidyashree K P4
1,2,3 Final year student, Dept. of Information Science and Engineering, Vidyavardhaka College of Engineering,
Mysuru, India,
4 Assistant Professor, Dept. of Information Science and Engineering, Vidyavardhaka College of Engineering ,
Mysuru, India,
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - It's been really tough to teach students on an
online platform and make the lesson interesting during the
COVID-19 pandemic. As a result, there was a need for a dust-
free classroom for children. Using MediaPipe andOpenCV, this
article presents a unique paintapplicationthatidentifieshand
movements and tracks hand joints. This program uses hand
gestures to give users with an intuitive method of Human
Computer Interaction (HCI). HCI's major goal is to improve
human-computer interaction.
Key Words: Hand Gesture Recognition, Human Computer
Interaction, Computer Vision, Paint, MediaPipe
1. INTRODUCTION
Currently, people and machines interact mostly through
direct contact methods such as the mouse,keyboard, remote
control, touch screen, and other similar devices, whereas
people communicate primarilythroughnatural andintuitive
non-contact methods such as sound and physical motions.
Many researchers have attempted to help the computer
identify other intentions and information using non-contact
methods like as voice, facial expressions, physical motions,
and gestures. A gesture is the most crucial aspect of mortal
language, and gestures also play a significant role in mortal
communication. They're considered to be the simplest way
for humans and computers to communicate. Sign language
recognition, robotics, and other applications fall under the
category of gesture recognition.
Two methods are typically used for Gesture recognition
for HCI applications. The first is based on wearable or direct
physical approaches, while the second is based on computer
vision and does not require any sensors to be worn. The
data-glove, which is made up of sensors to capture hand
motion and location, is used inthewearableordirect contact
technique. The vision-based technique uses the camera to
provide contactless communication between humans and
machines. Computer vision cameras are simple to operate
and inexpensive. However, due to differences in hand sizes,
hand position, hand orientation, lighting conditions, and
other factors, this approach has certain limitations.
In this paper, weintroduce avirtualpaintapplicationthat
uses hand gestures for real-time drawing or sketching onthe
canvas. Hand gesture-based paint application can be
implemented using cameras to capture hand movement. To
accomplish activities like as tool selection, writing on the
canvas, and clearing the canvas, an intangible interface is
created and implemented using vision-based real-time
dynamic hand gestures. The images of the hands are taken
with the system's web camera and processed in real time
with a single-shot detector model and media pipe, allowing
the machine to communicate with its user in a fraction of a
second.
2. LITERATURE REVIEW
Many methods are used for hand gesture recognition in
real-time. Sayem Mohammad Siam, Jahidul AdnanSakel, and
Md. Hasanul Kabir has proposed a new method of HCI
(Human-Computer Interaction), that uses marker detection
and tracking technique. Instead of having a mouse or
touchpad, two colored markers are worn on the tips of the
fingers to generate eight hand movements to provide
instructions to a desktop or laptop computer with a
consumer-grade camera [1]. They have also used the
"Template matching" algorithm for the detectionofmarkers
and Kalman Filter for tracking. In [2] the developed system
uses a data glove-based approach to recognize real-time
dynamic hand gestures. The data glove has ten soft sensors
integrated in it that measure the joint angles of five fingers
and are used to collect gesture data. Real-time gestures are
recognized using techniques such as gesture spotting,
gesture sequence simplification, and gesture recognition.
Shomi Khan, M. Elieas Ali, Sree Sourav Das have
developed a system that uses a skin color detection
algorithm to convert ASL (AmericanSignLanguage)intotext
from real-time video [3]. Because skin color and hand shape
differ from person to person, detecting the hand might be
challenging. The technology uses two neural networks to
overcome this. The SCD (Scalable color descriptor) neural
network is the first. The picture pixels are fed into the SCD
neural network, which determines whether or not they are
skin pixels. The second one is HGR (Hand gesture
recognition) neural network to which the extractedfeatures
will be provided. The features will be extracted by two
distinct algorithms namely Finding the fingertip and Pixel
segmentation.
Pavitra Ramasamy and Prabhu G [4] have proposed a
revolutionary technology in which the user can write the
alphabet or type whatever he orshe wantsbymerelywaving

his or her finger over a colorful LED light source. Only the
color of the LED is tracked to extract the movement of the
finger sketching the alphabet. The color ofthetrackedobject
is changed to white, while the background is changed to
black. The black and white frames are stitched together to
create a single black and white imageofthealphabetthat the
user wanted to draw.
[5] To accomplish mouse actions such as moving the
mouse cursor, clicking left, and clicking right with hand
gestures, an intangible interface is conceived and
implemented utilizing vision-based real-time dynamichand
gestures. MATLAB is used for the implementation of the
system. S. Belgamwar and S. Agrawal [6] have developed a
new HCI technique that incorporates a camera, an
accelerometer, a pair of Arduino microcontrollers, and
Ultrasonic Distance Sensors. The primary concept behind
this interface is to capture motionsusingUltrasonic Distance
Sensors. The distance between the hand and the distance
sensor is calculated to record the gestures.
For 3D hand gesturedetection, Quentin De Smedt,Hazem
Wannous, and Jean-PhilippeVandeborre [8]usedaskeleton-
based model. They used the geometric shape of the hand to
obtain an effective descriptorfromtheIntelReal-Sensedepth
camera's hand skeleton linked joints. The skeleton-based
approach is better than the depth-based approach. In [9]
Prajakta Vidhate, Revati Khadse, and Saina Rasal have
developed a virtual paint application that uses ball-tracking
technology to track the hand gestures and write on the
screen. They have used a glove with a ping pongballattached
to it asa contour. [10] RuiminLyu, Yuefeng Ze, WeiChen,and
Fei Chen presented a customizable airbrush model that uses
the Leap Motion Controller, which can track hands, to create
an immersive freehand painting experience.
3. ALGORITHM USED FOR HAND TRACKING
Hand gesture recognition and tracking are handled bythe
MediaPipe framework, while computer vision is handled by
the OpenCV library. To track andrecognizehandmovements
and hand tips, the program makes use of machine learning
ideas.
3.1 MediaPipe
MediaPipe is a Google open-source framework thatwas
initially released in 2019. MediaPipe has some built-in
computer vision and machine learning capabilities. A
machine learning inference pipeline is implemented using
MediaPipe. ML inference is the process of running real data
points. The MediaPipe framework is used to solve AI
challenges that mostly include video and audio streaming.
MediaPipe is multimodal and platform independent. As a
result, cross-platform apps arecreatedusingtheframework.
Face detection, multi-hand tracking, hair segmentation,
object detection, and tracking are just a few of the
applications that MediaPipe has to offer. MediaPipe is a
framework with a high level of fidelity. Low latency
performance is provided throughtheMediaPipeframework.
It's in charge of synchronizing time-series data.
The MediaPipe framework has been used to design and
analyze systems using graphs, as well as to develop systems
for application purposes. In the pipeline configuration, all of
the system's steps are carried out. The pipeline that was
designed can run on a variety of platforms and can scale
across desktops and mobile devices. Performance
evaluation, sensor data retrieval, and a collection of
components are all part of the MediaPipe framework.
Calculators are the parts of the system. The MediaPipe
framework uses a single-shot detector model for real-time
detection and recognition of a hand or palm. Itisfirsttrained
for the palm detection model in the hand detection module
since palms are easier to train. It designates a hand
landmark in the hand region, consisting of 21 joint or
knuckle coordinates as shown in the Figure 1.
Fig -1: Coordinates or landmarks in the hand
3.2 OpenCV
The computer vision library OpenCV is a must-have for
everyone who works with computers. It includes object
detection image-processing methods. OpenCV is a python
package for creating real-time computer visionapplications.
Image and video processing and analysis are handled by the
OpenCV library.

4. ALGORITHM USED FOR HAND TRACKING
The various constraints in the system are explained in the
flowchart of the Virtual Paint Application in Figure 2.
Fig -2: Flowchart of the virtual Paint Application
The virtual paint application presented is based on the
frames recorded by the PC's web camera. The frames are
captured by the web camera and sent to the system. The
method makes use of a web camera tocaptureeachframetill
the application is finished. Thevideoframesaretransformed
from BGR to RGB color to locate the hands in the video
frame. The system then determines which finger is up by
comparing the tip Id of the corresponding finger found via
the MediaPipe to the respective coordinates of the up
fingers, and then performs the appropriate function. The
user can write anything on the screen if his or her index
finger is raised. If both index finger and middle fingerare up,
the user can either change position on the screen or can
select any tool provided in the toolbar of the application. If
all the fingers are up except the thumb finger, the user can
clear the screen. If all the fingers are up, then no action is
performed on the screen.
5. CONCLUSION
The virtual paint application's fundamental goal is to
deliver an AI-based tool that allows users to draw anything
on screen using hand movements. This system alsogivesthe
user the option of selecting any tool from the toolbar. The
user can save their completed work or see their drawing
process as a replay animation with this application.
6. FUTURE WORK
This work can be further improved by experimentingwith
different interpolation methods such as PyGame which
includes a line drawing method that could help produce
smoother and cleaner lines. In the same vein, a variety of
brush shapes and textures can be implemented to make this
application more robust.
REFERENCES
[1] Siam, Sayem & Sakel, Jahidul & Kabir, Md. . Human
Computer InteractionUsingMarkerBasedHandGesture
Recognition., 2016
[2] M. Lee and J. Bae, "Deep Learning Based Real-Time
Recognition of Dynamic Finger Gestures Using a Data
Glove," in IEEE Access, vol. 8, pp. 219923-219933,2020,
doi: 10.1109/ACCESS.2020.3039401.
[3] S. Khan, M. E. Ali, S. Das and M. M. Rahman, "Real Time
Hand Gesture Recognition by Skin Color Detection for
American Sign Language," 2019 4th International
Conference on Electrical Information and
Communication Technology (EICT), 2019, pp. 1-6, doi:
10.1109/EICT48899.2019.9068809.
[4] P. Ramasamy, G. Prabhu and R. Srinivasan, "An
economical air writing system converting finger
movements to text using web camera," 2016
International Conference on Recent Trends in
Information Technology (ICRTIT), 2016, pp. 1-6, doi:
10.1109/ICRTIT.2016.7569563.
[5] P. Ramasamy, G. Prabhu and R. Srinivasan, "An
economical air writing system converting finger
movements to text using web camera," 2016
International Conference on Recent Trends in
Information Technology (ICRTIT), 2016, pp. 1-6, doi:
10.1109/ICRTIT.2016.7569563.
[6] S. Belgamwar and S. Agrawal, "An Arduino Based
GestureControl SystemforHuman-ComputerInterface,"
2018 Fourth International Conference on Computing
Communication Control and Automation (ICCUBEA),
2018, pp. 1-3, doi: 10.1109/ICCUBEA.2018.8697673.

[7] Y. Wu and C. -M. Wang, "Applying hand gesture
recognition and joint tracking to a TV controller using
CNN and Convolutional Pose Machine," 2018 24th
International Conferenceon PatternRecognition(ICPR),
2018, pp. 3086-3091,doi:10.1109/ICPR.2018.8546209.
[8] Q. De Smedt, H. Wannous and J. Vandeborre, "Skeleton-
Based Dynamic Hand Gesture Recognition," 2016 IEEE
Conference on ComputerVisionandPatternRecognition
Workshops (CVPRW), 2016, pp. 1206-1214, doi:
10.1109/CVPRW.2016.153.
[9] Prajakta Vidhate, Revati Khadse, Saina Rasal, “Virtual
paint application by hand gesture recognition system”,
2019 International Journal of Technical Research and
Applications, 2019, pp. 36-39.
[10] R. Lyu et al., "A flexible finger-mounted airbrush model
for immersive freehand painting," 2017IEEE/ACIS16th
International Conference on Computer and Information
Science (ICIS), 2017, pp. 395-400,
doi:10.1109/ICIS.2017.7960025.

VIRTUAL PAINT APPLICATION USING HAND GESTURES

More Related Content

What's hot (20)

Similar to VIRTUAL PAINT APPLICATION USING HAND GESTURES (20)

More from IRJET Journal (20)

Recently uploaded (20)

VIRTUAL PAINT APPLICATION USING HAND GESTURES