← Back to cohort

Ikram Ullah Khan

FAST · 2022 · i18-0743

—

Phone

—

GitHub

—

Academic

Program

BSCS

CGPA

—

Year

2022

Education

SEECS

Address

—

DOB

—

Career

Current role

—

Target role

—

Skills

MERN Stack, AWS, Raspberry Pi, Amazon Sagemaker, TensorFlow, Python, React, Node.js, Docker, Kubernetes, GitHub

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.

RecogLip – Sentence Level Lip Reading
The main motivation of this project is to help the people suffering from diseases like Amyotrophic lateral sclerosis (ALS), Aphasia and Aphonia. The patients suffering from such diseases can move their lips but unfortunately they cannot speak and we cannot listen to them. In order to overcome this issue, we developed a headgear using Raspberry Pie, that will detect the movement of lips and the text will be generated on the screen. Also, the voice will be generated by sound module. In this way, our headgear will help the patients to communicate.
Features include:
1. Detection of Lips
2. Generate Text
3. Generate Voice
4. Communication
RecogLip
Sentence Level Lip Reading Headgear
Group Members:
Ikram Ullah Khan (i18-0743) Usama Khan (i19-0551) Sobia Noor (i19-0744)
Supervisor: Dr. Hammad Majeed
Co-Supervisor: Mr. Saad Salman
OBJECTIVE
A headgear that will help people suffering with certain speech related diseases to communicate. A camera module connected with Raspberry Pi will detect the speech from the movements of lips and a speaker will be used to communicate that output.
ARCHITECTURE
REAL-TIME VIDEO STREAM FROM HEADGEAR
FACE MESH TO DETECT FACE POINTS
EXTRACT USEFUL FEATURES
AUDIO GENERATED BY SPEAKER MODULE
VIDEO TO TEXT
PASSING TO MODEL
TIMELINE
SEP OCT
PHASE 1
• A WEB APPLICATION FOR DATA COLLECTION
• PREPROCESSING THE DATA
• WORKING ON MODEL
• BUILDING A RASPBERRY PI HEADGEAR PROTOTYPE
NOV DEC
PHASE 2
• USING IMAGE PROCESSING TO EXTRACT BETTER FEATURES
• TRAINING THE MODEL
• OPTIMIZING THE PERFORMANCE AND ACCURACY OF THE MODEL
JAN MAR
PHASE 3
• RETRAIN THE TUNED MODEL
• WORKING ON CI/CD PIPELINE
• TESTING ON REAL-TIME
APR MAY
PHASE 4
• INTEGRATING MODEL THROUGH API'S
• BUILDING CI/CD PIPELINE
• WORKING ON BONUS FEATURES
TOOLS
Technology Used:
MERN Stack, AWS, Raspberry Pie, Amazon Sagemaker, TensorFlow
Supervisor Name:
Dr. Hammad Majeed
Group Members:
Usama Khan (i19 - 0551)
Sobia Noor (i19 - 0744)
Ikram Ullah Khan (i18 - 0743)

AI enrichment

Ikram Ullah Khan is a BSCS graduate who contributed to a final year project developing a Raspberry Pi-based headgear for sentence-level lip reading to assist patients with speech impairments. The project involved building a MERN stack web application, training TensorFlow models on AWS SageMaker, and implementing real-time video processing and text-to-speech features.

Skills (AI)

["MERN Stack", "TensorFlow", "AWS", "Amazon SageMaker", "Raspberry Pi", "Computer Vision", "Image Processing", "API Integration", "CI/CD Pipeline"]

Status: ai_done

Provenance

Source file: FAST - School of Computing -Graduate Directory-2023.pdf
From job #14 page 336
Created: 1778169248