You are on page 1of 22

Visvesvaraya Technological University

Belagavi, Karnataka

An Internship Seminar Presentation


On
“Deep Learning using PyTorch for Fall Detection”

Presented By

ARJUN C M 4JN20CS013

Under the Guidance of : Coordinator :


Mrs Thaseen Bhashith B.E., M.Tech Mrs Sreedevi S B.E., M.Tech
Assistant Professor, Assistant Professor,
Dept. of CS&E Dept. of CS&E
Contents
 Company Profile
 Initiatives of CIE
 Abstract
 Technical skills
 Task Performed
 Results
 Reflection Notes
 Conclusion
 References
Company Profile

The internship was carried at Jawaharlal Nehru New College of Engineering, Shivamogga. It
was carried out under JNNCE EI LEARN Platform – a collaboration between JNN
College of Engineering with Ekathva Innovations Pvt. Ltd.
 Ekathva Innovations was established in 2018 in Karnataka at the campus of JNNCE.
 Ekathva innovations mainly aims in making user friendly and economic products in the areas of
design and development of the Embedded Systems.
 The company also aims at web development and app development both on android and IOS in
the field of information technology.
 They are experienced in providing high quality solutions to clients in diverse business areas.
Center for Innovation & Entrepreneurship ( CIE)

 Center for Innovation and Entrepreneurship (CIE) was started in 2018 in JNNCE with the
support of various Government departments and agencies.

 CIE aims to create innovation, startup and entrepreneurial ecosystem at JNNCE in &
around Shivamogga.

 It provides all the required facilities like incubation support, funding and infrastructure
space for startup establishment
Initiatives of CIE
 New Generation Innovation and Entrepreneurship Development Centre

 New Age Incubation Network (NAIN) Centre

 JNNCE Research & Innovation Centre (JNNRIC)

 JNNCART

 Entrepreneurship Cell (E-Cell)

 Technology Business Incubator (TBI)

 Technology Activities & Programs (TAP)


Abstract
The internship as a intern for Ekathva innovations has been the most rewarding and motivational
experiences. Fall detection is a critical aspect of ensuring the safety and well-being of
individuals, particularly in contexts such as eldercare and healthcare facilities. The project begins
with a comprehensive exploration of existing fall detection methods and the challenges they
pose. Traditional methods often suffer from limitations in accuracy and scalability, motivating the
adoption of deep learning techniques facilitated by frameworks like PyTorch.
Technical skills

 Overview of AI

 NumPy

 PyTorch

 Convolutional Neural Networks


Overview of AI
• AIML stands for Artificial Intelligence Modelling Language. AIML is an XML based markup
language meant to create artificial intelligent applications.
• AIML makes it possible to create human interfaces while keeping the implementation simple to
program, easy to understand and highly maintainable.

Credentials of AI
NumPy

Benefits of NumPy:
More speed: NumPy uses algorithms written in C that complete in nanoseconds rather than
seconds.
 Fewer loops: NumPy helps to reduce loops and keep from getting tangled up in iteration
indices.
 Clearer code: Without loops, code will look more like the equations trying to calculate.
 Better quality: There are thousands of contributors working to keep NumPy fast, friendly,
and bug free.
PyTorch

 Pythonic Nature: Python is one of the fastest growing programming languages from
the last 5-10 years.

 Easy to Learn.

 Strong Community.

 Easy Debugging.
Convolutional Neural Networks
Convolutional Neural Networks (CNNs) are a powerful tool for machine learning, especially
in tasks related to computer vision.
Convolutional Neural Networks, or CNNs, are a specialized class of neural networks designed to
effectively process grid-like data, such as images.

Benefits of CNN:
 No require human supervision required.
 Automatic feature extraction.
 Highly accurate at image recognition & classification.
 Minimizes computation.
 Ability to handle large datasets.
CNN Architecture
 Convolutional layers apply filters to the input image, which effectively detect features like
edges, textures, or patterns. Each filter learns to detect a specific feature by convolving it
across the input image and producing a feature map.
 Maxpool2D divides the input into a grid and computes the maximum value for each grid
cell
 After the convolutional and pooling layers extract features from the input images, Flatten
rearranges these features into a single continuous vector.
 The model is evaluated and various graphs are plotted to show accuracy.
 To show the prediction, the model is fed with a image and its label is predicted.
Task Performed

 Implemented by using NumPy, Pytorch and its operations on sets of data for efficient
manipulation and analysis.
 Dataset is split into training dataset and testing dataset.
 Defining a custom neural network model (FDNet) based on the MobileNetV2 architecture
 Convolution Neural Network(CNN) is used to train the model.
 Computing the accuracy of the model.
 Displaying the confusion matrix with labeled axes and values.
 Visualize the Final Output by using Plotting Library.
Results

The AI-Based Fall Detection System project aims to develop an automated system using
machine learning algorithms to detect and respond to falls in real-time. The project involves
preprocessing the data to extract relevant features, selecting and training machine learning
models, evaluating the performance of the models, integrating them into a real-time system,
testing and validating the system in real-world scenarios. The outcome is a reliable and
automated solution for detecting falls, enhancing the safety and well-being of individuals at
risk of falls, especially in elderly care and healthcare settings.
Accuracy of The Model

Snapshot of Accuracy of model


Output of The Model

Snapshot of Output of Result


Visualization of The Output
Reflection Notes

This internship provided an opportunity to carry out a real time mini project. The internship
training gave an opportunity to express individual’s ideas and to put those ideas in the
project assigned. My internship as a intern for Ekathva innovations has been the most
rewarding and motivational experiences. In this Internship, I got to learn so many
frameworks and modern world technologies which are prominently used for projects around
the world and this is definitely a valuable asset for my upcoming career as a software
engineer.
Conclusion

The internship on AI for structured data has provided a comprehensive stage for learning
fundamental operations in NumPy to implementing advanced techniques using PyTorch for
predictive modeling. This project, developed a deep learning model using convolutional neural
networks (CNNs) to classify images. The collected dataset of images and trained our model to
recognize different categories or classes within the images. By optimizing the model's parameters
and evaluating its performance, achieved a satisfactory accuracy level. Our trained model can now
accurately classify new images into their respective categories, demonstrating the effectiveness of
deep learning techniques for image classification tasks.
Reference

[1] https://numpy.org
[2] https://pytorch.org
[3] https://numpy.org/learn
[4] https://web.mit.edu/dvp/Public/numpybook.pdf
[5] https://github.com/nithiroj/Fall-Detection-PyTorch.git
THANK YOU

You might also like