http://hdl.handle.net/123456789/98 2026-08-01T18:18:30Z 2026-08-01T18:18:30Z Yasir, Shiekh Reg # 78981 Najam, Adeel Reg # 79240 Usman, Muhammad Reg # 78997 http://hdl.handle.net/123456789/21503 2026-07-15T05:48:20Z 2025-01-01T00:00:00Z

VOGUEAI: THE NEXT-GEN VIRTUAL STYLIST Yasir, Shiekh Reg # 78981; Najam, Adeel Reg # 79240; Usman, Muhammad Reg # 78997 The research aims to develop VogueAI as a web-based smart virtual assistant that provides users with virtual clothing trials before they make purchases. People find online shopping difficult because they are unsure if products will match their body and feature that enables users to fit properly. VogueAI addresses this problem through a submit their personal images that generate how various outfits appear on their physique. techniques the system applies digital outfits to user-supplied Via computer vision images for results which closely represent actual garments. Users get outfit recommendations through an integrated chatbot feature of the platform. From your event type to your current mood, you can seek advice through the system . Users can find their fashion inspiration through authentic clothing by making a request brands Rastah, Nishat, and ELO which keep their fashion collection contemporary and relevant to buyers. web technologies to create its platform. The platform . Users The system implements optimizes itself to operate efficiently on mobile phones and desktop computers free basic features but premium features including high resolution try have access to and personalized outfit combinations require an upgrade to the premium version. ons We intend to enhance our virtual fitting tool by developing improved body-type recognition algorithms while creating more realistic fabric textures for users in future future development includes viewing clothing from multiple angles and updates. Our developing advanced styling features and eco-awareness labels to show environmental reach additional regions globally while adding multi footprint of garments. We plan to language capabilities to our platform to receive friend recommendations before making purchases. and build social sharing tools which would allow users VogueAI brings together fashion an improved custom online shopping experience. elements with technological capabilities to deliver Supervised by Saghir Ahmed

2025-01-01T00:00:00Z Ahmed, Zaheer Reg # 79244 Ahmed, Rana Irtaza Reg # 78996 Tauqeer, Syeda Marium Reg # 79000 http://hdl.handle.net/123456789/21504 2026-07-15T05:50:20Z 2025-01-01T00:00:00Z

QUICKQUIZ:AI DRIVEN ALGEBRA QUIZ GENERATOR USING LLM INTEGRATION Ahmed, Zaheer Reg # 79244; Ahmed, Rana Irtaza Reg # 78996; Tauqeer, Syeda Marium Reg # 79000 Typical practice tools are not flexible and keep offering the same problems, not changing with the student’s results or what they need help with. While many people are using e-learning nowadays, there are still not many options for Ai generated and customized math quizzes focused on algebra. Making quizzes the conventional way is tedious and usually left unchanged, making it hard for teachers and students to update content based on their skills and subjects. In addition, basic question banks do not evaluate understanding in real time or provide enough variety for students to be challenged. As a result, QU1CKQUIZ was developed as an Al-based system to create and assess quizzes in algebra on its own. The primary goal is to make math practice more engaging, available to all and simpler for everyone involved in learning. As an important topic in math, algebra needs consistent practice and QUICKQU1Z provides this by generating intelligent exercises automatically. The system was built using Flask on the backend which gave users a fast and interactive web page. An LLM was included to build algebra quizzes in three types: MCQs, Fill-in-the-Blanks and True/False questions. To maintain topic relevance, prompt engineering was applied to remove anything that did not involve mathematics. The system checks how users react quickly and helps them with instant feedback. Users were given a new feature making it easier to see how accurate GPT’s generated content was. As a result, users can interact with the program, their data is managed checked on the backend to give them a smooth and The system was released as a web application which proved properly, and feedback is informative experience, its value in creating quizzes, checking user inputs and engaging users Supervised by Muhammad Shahid Khan

2025-01-01T00:00:00Z Adnan, Muznah Reg # 79902 Waseem, Abdul Hadi Reg # 79258 http://hdl.handle.net/123456789/21500 2026-07-15T05:04:18Z 2025-01-01T00:00:00Z

A DEEP-LEARNING AND IOT-BASED AUTOMATED CLASSIFICATION OF LUNG DISEASES Adnan, Muznah Reg # 79902; Waseem, Abdul Hadi Reg # 79258 In 2020, respiratory diseases caused over 86 thousand deaths in Pakistan, ranking the country 8th in the world for this health hazard. The majority of the population does have access to adequate healthcare facilities, including the 61.8% of Pakistanis who live in luial aieas. The combination of a lack of infrastructure, health literacy severely limits the healthcare options for necessitates the development of intelligent systems that are able to restructure responsive and timely healthcare approaches for rural alleviation of respiratory diseases. not trained personnel, and most citizens. This situation communities, aiding in the To develop improved healthcare facilities, piopose the integration of deep learning we and the Intel net for advanced remote caregiving solutions. Oui proposed solution achieved exceptional performance with 95.05% the Asthma Detection Dataset Version 2[10], comprising 1,211 lung sound samples across 5 respiratory conditions. The system is based around three components: The IoT smart stethoscope, the deep learning-model-based data classification achieving clinical-grade performance, and the analysis and visualization web application. accuracy on With the inclusion of these components, the goal of our system is to transform respiratory healthcare in Pakistan, transformation especially for the rural population, lequires the responsible development of sophisticated devices while concurrently enhancing individual health seek to This diagnostic management capabilities. We mitigate the accessibility and knowledge gap between patients and health practitioners using modern technologies combined with simple-to-use applications. Supervised by Dr. Taha Jilani

2025-01-01T00:00:00Z Riaz, Muhammad Haris Reg # 79013 Azmeen, Ramisha Reg # 79008 Abdullah, Muhammad Reg # 79021 http://hdl.handle.net/123456789/21496 2026-07-15T04:56:14Z 2025-01-01T00:00:00Z

CLEARZONE - FOCUSES ON CREATNG CLEAN, WASTE-FREE ZONES IN COMMUNITIES Riaz, Muhammad Haris Reg # 79013; Azmeen, Ramisha Reg # 79008; Abdullah, Muhammad Reg # 79021 The ClearZone project aims to create a useful waste management platform that bridges the gap between individuals and NGOs, fostering collaborative efforts for a cleaner environment. The platform integrates geolocation-based waste reporting, an NGO dashboard, user leaderboards, and a notification system to streamline waste management processes. Significant progress has been made in deploying the waste classification model on Hugging Face, where it is publicly available. However, the integration of the model into the web application remains pending. The mapping feature, which visualizes waste locations, has been partially implemented and currently functions on one page. The notification system is yet to be fully developed and integrated, which will enhance user and NGO engagement by providing real-time updates on waste report statuses. Key completed features include user and NGO authentication, profile management, waste reporting (without CNN-based auto-classification), and a partially functional geolocation module. Upcoming milestones involve finalizing the integration of the deployed model into the web app, expanding the mapping functionality to all relevant pages, and completing the notification module to provide a seamless user experience. The project's overall structure and progress indicate its potential to significantly impact environmental sustainability by simplifying waste reporting and management. Once the pending tasks are completed, ClearZone will be a fully functional platform ready for real-world application. Supervised by Fatima Zafar

2025-01-01T00:00:00Z