← Back to cohort

Talha Ali Khan

FAST · 2023 · 19I-1681

—

Phone

—

GitHub

—

Academic

Program

BSCS

CGPA

—

Year

2023

Education

SEECS

Address

—

DOB

—

Career

Current role

—

Target role

—

Skills

Python, PyTorch, Flutter, MongoDB, Firebase, BeautifulSoup, Selenium

Verbatim text

The exact text the LLM saw on the page (or the booklet text from the old import). This is what powers semantic search.

DressEZ
A cross-platform mobile app made in Flutter, that utilizes the power of deep learning to recommend clothes to its customers. It uses the deep learning model ResNet50, transfer learned for multi-label classification, to generate embeddings for images of clothes. The customer uses the app to take a picture of one of their clothes, and the model generates an embedding for the image. The app calculates the cosine similarity of this embedding with the embeddings of the clothes in its database. The most similar looking clothes along with the links to where the customer can buy them are returned as the recommendations. The clothes in the app's database are scrapped from local online stores.
App Features:
• A user can take pictures of their clothes from within the app. The pictures of clothes taken by a user are stored online in a Firebase database.
• The camera screen in the app has an overlay of a t-shirt or of a pair of pants that help guide the user in taking a picture.
• A user can select one of their images from within to app to generate recommendations. The recommendations are generated and displayed to the user.

DressEZ
AN APP THAT ACTS AS A PERSONAL STYLIST FOR ITS USER
SUPERVISOR: DR. FAISAL CHEEMA
TEAM:
ALI MURTAZA 19I-1665
TALHA ALI 19I-1681
SAAD JAFAR 19I-1691

OBJECTIVES
RECOMMENDS WHAT CLOTHES TO BUY AND FROM WHERE, AND ALSO SUGGESTS WHAT COMBINATION OF CLOTHES TO WEAR FOR EACH DAY

ARCHITECTURE
PICTURES OF CLOTHES -> MOBILE APP -> RECOMMENDATION MODEL -> RECOMMENDED ITEMS / LINK TO ITEMS

TOOLS
python Selenium BeautifulSoup Flutter PyTorch

TIMELINE
SEP: COLLECTING AND MANUALLY CLEANING DATA FROM AVAILABLE DATASETS
OCT: PREPROCESSING DATA AND TRAINING THE NEURAL NETWORK MODEL
NOV-DEC: CREATING A FRONTEND IN THE FORM OF A MOBILE APPLICATION
JAN-FEB: CREATING A DATABASE OF CLOTHES FROM WEB SCRAPING FROM LOCAL ONLINE STORES
MAR: INTEGRATING THE MODEL AND THE DATABASE INTO MOBILE APPLICATION
APR: OPTIMIZING PERFORMANCE AND DEPLOYING MODEL ON A SERVER

Technology Used:
Python, PyTorch, Flutter, MongoDB, Firebase, BeautifulSoup, Selenium
Supervisor Name:
Dr. Faisal Cheema
Group Members:
Ali Murtaza (19I-1665)
Talha Ali Khan (19I-1681)
Saad Jafar (19I-1691)

AI enrichment

Talha Ali Khan is a BSCS graduate who developed DressEZ, a cross-platform Flutter mobile application that uses a transfer-learned ResNet50 model for clothing recommendations. The project involved web scraping data, training deep learning models with PyTorch, and integrating them with Firebase and MongoDB backends.

Skills (AI)

["Flutter", "Python", "PyTorch", "ResNet50", "Deep Learning", "Firebase", "MongoDB", "Web Scraping", "Selenium", "BeautifulSoup", "Computer Vision"]

Status: ai_done

Provenance

Source file: FAST - School of Computing -Graduate Directory-2023.pdf
From job #14 page 499
Created: 1778169842