Pratyush Garg

Save Child Bharat

This project is a dynamic real-time child safety monitoring system designed to enhance domestic security for children aged 0–6 years. It integrates facial recognition to identify and track children across multiple camera feeds and employs YOLO-based object detection to identify hazardous objects (e.g., knives, scissors, bottles) within user-defined danger zones. The system fetches live configuration data from an API, enabling seamless addition, removal, or modification of users, cameras, and zones without requiring application restarts. It combines multiprocessing for user-level isolation and multithreading for efficient parallel camera monitoring, ensuring scalable and adaptable performance in real-world home environments.

Languages/Tools:

Python, OpenCV, Ultralytics YOLO, pickle, RTSP

RealEstate Website

FindProperty
FindProperty is a real estate platform, similar to UK-based sites like Rightmove and Zoopla, where users can buy, sell, and rent properties such as homes, flats, and commercial spaces. I contributed to the development of the platform, primarily using Angular for the front-end, which enabled dynamic, responsive user interfaces. The platform also incorporated DALL·E for generating interactive media, such as property images and videos, enhancing the overall user experience with dynamic and personalised content. The use of HTML5 and CSS3 ensured the platform’s responsiveness across multiple devices. As part of the team, I worked on implementing search functionalities and optimising the overall user journey, ensuring the platform aligned with business goals and provided a seamless experience for users.

LocalHouse
LocalHouse operates as a property agency platform, where properties are listed directly by the agency. The platform allows potential buyers and renters to connect with agents efficiently. For this project, I worked on creating a clean and functional interface using HTML5 and CSS3, ensuring easy navigation and smooth interactions between users and agents. The focus of this platform was on simplicity and direct communication, allowing clients to quickly find the necessary property details and contact agents. My role involved enhancing the user experience, ensuring that the platform was intuitive and easy to navigate.

SmartRent
SmartRent is a platform that functions similarly to Airbnb, enabling users to book short-term rental properties. The platform is designed to provide an efficient and user-friendly experience for both guests and hosts. I contributed to the project by developing the front-end using HTML5 and CSS3, ensuring the site was visually appealing and responsive. The platform’s primary features include property listing management, booking systems, and payment processing. I focused on ensuring smooth navigation and quick access to necessary information, allowing users to easily book or list properties. The project required effective collaboration with cross-functional teams to deliver a robust and reliable platform.

Languages/Tools:

HTML5, CSS3, Angular, DALL·E, RumwayMl

WhatsApp Broadcast Bot

The project titled "Transforming Social Media: Development of a Multi-Platform Broadcast Bot" focuses on creating an innovative solution for seamless communication across multiple social media platforms. By leveraging browser automation, Python, and natural language processing (NLP), the bot is designed to send media-rich content, such as text, images, and videos, to diverse audiences efficiently.

The project addresses key challenges such as bulk messaging, real-time translation, and cross-platform compatibility, allowing users to interact across platforms like WhatsApp, Telegram, and Instagram. It integrates technologies like Playwright for browser automation and Argos Translate for language translation, ensuring user-friendly communication regardless of language barriers.

The bot's modular and scalable design also provides robust error handling, making it a reliable tool for both personal and business use, enhancing modern digital communication practices.lex data in an accessible and visually engaging manner, enabling users to glean insights effortlessly.

The Repo is private, email for full source code.

Languages/Tools:

Python, Streamlit, Playwright, ArgosTechTranslate

Streamlit Dashboards

This initiative involves the development of advanced data visualization and analytics dashboards, akin to PowerBI in functionality and design. The dashboards feature interactive graphs and user-friendly interfaces, making them highly effective for data interpretation and decision-making. A key aspect of this project is the integration and analysis of YouTube live stream data, which is utilized to create these comprehensive dashboards. The aim is to present complex data in an accessible and visually engaging manner, enabling users to glean insights effortlessly.

The Repo is private, email for full source code.

Languages/Tools:

Python, Streamlit

AI Content Innovation

This comprehensive project showcases an extensive application of AI in various realms of content creation and development. Key components include enhancing ChatGPT 4 functionalities, developing AI avatars and images, utilizing illusion diffusion models for sophisticated visual effects, and converting text into diverse media formats. Additionally, the project encompasses the creation and management of faceless YouTube channels, AI speaking avatars, and an AI news anchor, demonstrating a pioneering approach in digital media. These efforts highlight the fusion of AI with creative content strategies, producing high-quality, engaging, and innovative content for a range of digital platforms.

Languages/Tools:

Adobe Premier Pro, CapCut, Dalle 3, Pika Labs, ChatGPT 4 + (Bots), Midjourney, Runway ML, LeonardoAi, Huggingface.io, Narakeet, Elevenlabs, Haygen, Wav2Lip.

Digital Business Card

This project involves the creation of dynamic e-business cards, revolutionizing traditional networking tools with a digital twist. Moving beyond standard designs, these business cards are featured as hosted e-cards with animated QR codes. Users can scan these codes to directly access the card's details online. Each QR code is uniquely tailored to reflect the business or individual it represents, offering a creative and personalized approach. This innovative method not only modernizes the business card format but also significantly enhances user interaction and engagement, creating a lasting impression in professional exchanges.

Languages/Tools:

HTML5, CSS3, JavaScript, Huggingface.co (for QR codes).

Research Analysis

Cholesterol levels show how much cholesterol is circulating in your blood. A drastic change in cholesterol levels leads to heart disease.Cigarette smoking, high blood pressure, and high blood cholesterol are the most clearly established risk factors that have been identified as being strongly associated with coronary heart disease. We are currently analysing how the cholesterol level is different for males and females.

Dataset has 303 observations (rows) and 17 variables (columns). This study attempts to evaluate whether cholesterol level is different for male and female. We ask the following research question:

Is there a difference in the mean Cholesterol level in Cleveland between Male and Female populations?

The null hypothesis is: H0: There is no difference in the mean cholesterol level in Cleveland between male and female populations.

Alternative hypothesis is: H1: There is a difference in the mean cholesterol level in Cleveland between the male and female populations.

Firstly dataset is filtered for necessary outliers. Then from the newly created csv file visualization had been done to get a clear overview:-
1. Is are research question correct according to the data.
2. Which analysis techique to be used for the final result.

Languages/Tools:

R

Report Summary Generation

The focus of the project is on Extractive Summary Generation. The system allows users to input files in various formats, such as .pdf, .docs, or .txt. After receiving the input file, the system processes and cleans it to extract relevant sentences. To determine the importance of each sentence, a scoring mechanism utilizing a matrix is employed. The system is designed and deployed using PyQt5, providing a user-friendly graphical interface for inputting files and viewing the generated summaries. This project aims to create an efficient and effective way of generating concise summaries from text files, making use of modern technologies for enhanced user experience and accessibility.

When we initially created this application, there were only a few online apps available, and most of them were console-based. In contrast, we developed a comprehensive UI-based windows application using Python designer. This application offers several advantages over its counterparts:
1. It allows users to work seamlessly even when internet services are down, ensuring uninterrupted access and usability.
2. There is no need to open any website or remember URLs, making it more convenient and user-friendly.
3. Our application enables users to upload entire papers without any restriction on word count, making it highly versatile for different document sizes.
By developing this full UI-based windows application, we aimed to provide users with a more efficient and robust tool that enhances their workflow and simplifies document handling.

Languages/Tools:

Python, PyQT5

Reviewer

Introducing our groundbreaking web application – your ultimate destination for comprehensive product reviews across diverse market segments! Seamlessly blending cutting-edge technologies, our platform offers a seamless and interactive user experience. Imagine a space where you can explore product insights, all meticulously aggregated and updated in real-time. "REVIEWER"

Harnessing a multi-tiered approach, we've woven together HTML5, CSS, and JavaScript to craft an intuitive and visually captivating front-end. Our robust Back-End, powered by NodeJs, ensures flawless functionality and smooth navigation, enhancing your journey on our platform.

What sets us apart is our data aggregation prowess, thanks to Python's efficient web crawling capabilities. We meticulously extract vital product details from a myriad of online sources, providing you with a holistic overview that's continuously enriched with up-to-the-minute information.

But that's not all – we've gone above and beyond by integrating Selenium for web browser automation, expanding the horizon of data collection. Our platform spans three key product categories: Movies/TV Series sourced from IMDB, Mobile Phones curated from Gadgets NDTV, and Android Apps gleaned from the Google Play Store. This curated collection forms the heart of our extensive dataset, boasting an impressive range of content – 7 lakh movies and TV series from IMDB, 10,000 mobile phones from Gadgets NDTV, and 20,000 Android apps from the Google Play Store.

Efficiency is paramount, and that's why we've employed Python-generated CSV files to optimize data management. Accessing and manipulating information has never been more streamlined and convenient. Our journey doesn't stop at data collection; we've put extra effort into crafting a clean, cohesive dataset. Through meticulous data cleaning and precise formatting, we've ensured that the dataset maintains its accuracy and reliability.

Embark on this thrilling voyage of exploration and revelation. Immerse yourself in the seamless blend of cutting-edge technology and insightful data, crafting an unmatched user experience. As our platform unfolds, we extend a heartfelt invitation to discover a realm where comprehensive reviews are just a click away. Step into the forefront of product exploration – welcome to our web application! Your gateway to informed decisions and an enriched digital journey beckons.

Languages/Tools:

Python, HTML5, CSS, Selenium, Node.Js, Jupyter, Google Collab, Firebase

Movie Recommendation System

I have developed a console-based application utilizing Python, which offers real-time data updates. The application employs advanced data mining and web crawling techniques to extract live information from IMDB, ensuring users have access to the most current content. By seamlessly integrating Python's capabilities, the application guarantees efficient data collection and processing. This console-based solution allows users to interact seamlessly with real-time data feeds, enabling them to make informed decisions or stay updated with entertainment trends. The application's reliance on IMDB's live data calls ensures a dynamic and precise data stream, enhancing its reliability and accuracy.

The application comprises three distinctive sections, each catering to specific user needs. The "Search by Name" section enables users to input keywords such as movie titles, actors, or TV series names, ensuring efficient retrieval of pertinent information. The "Search by Actor" function appeals to users interested in exploring an actor's body of work, presenting comprehensive details of their filmography and contributions to the entertainment industry. The "Top Rated" section curates the most highly acclaimed content based on user reviews and ratings, guiding users toward quality entertainment options. This segmented structure facilitates effortless exploration, discovery, and access to desired content, contributing to a user-friendly experience.

Languages/Tools:

Python

School Management System

Basic school management system having records of student and also containing library database. Different Departments of school have different access level; Java, HTML, CSS.

Languages/Tools:

Java, Eclipse