Scribd
Upload
20 views12 pages

Project Presentation - 3 Template

The document outlines a project aimed at developing a machine learning-based system for Indian Sign Language (ISL) detection to aid the deaf community in India. It details the problem of communication barriers faced by hearing-impaired individuals and describes the implementation plan, including data collection, model training, and evaluation, with good accuracy achieved so far. Pending tasks include improving real-time detection accuracy and finalizing an IEEE paper on the project.

Uploaded by

deepmanera216
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
20 views12 pages

Project Presentation - 3 Template

The document outlines a project aimed at developing a machine learning-based system for Indian Sign Language (ISL) detection to aid the deaf community in India. It details the problem of communication barriers faced by hearing-impaired individuals and describes the implementation plan, including data collection, model training, and evaluation, with good accuracy achieved so far. Pending tasks include improving real-time detection accuracy and finalizing an IEEE paper on the project.

Uploaded by

deepmanera216
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

New Horizon Institute of Technology

and Management, Thane

Department of Artificial Intelligence and Data Science Engineering

Indian Sign Language Detection for Deaf


Community using Machine Learning

By
Saumya Darekar 12126001
Deep Manera 12116051
Nisha Patil 12116032

Guided by: Dr. Anu Malhan


Problem Statement

• In India, many people with hearing impairments use Indian Sign Language (ISL), but
limited understanding among the general public creates significant communication
barriers. This issue hampers access to services, education, and employment, while
current sign language interpretation methods are often impractical. There's a critical
need for a technology-driven solution to provide real-time, reliable translation of ISL
gestures into text.
Architecture(Block Diagram)
Implementation Plan (Gantt Chart)
Module Name

Execution of Completed Modules


● The code for data collection, training, and evaluation has been
implemented and tested.
● Model has been trained and evaluated, with results showing good
accuracy in testing.
Data collection

● Captures hand and face landmarks using MediaPipe and


stores gesture sequences.
● Live hand gestures
● Successfully recorded and stored landmark sequences for
training.
Model Training
● Trains LSTM + CNN model for gesture recognition.
● Preprocessed dataset
● Achieved good accuracy in evaluation.
Model Evaluation
● Assesses model performance using confusion matrix and classification reports.
● Identified misclassifications and accuracy trends.
Pending Modules

Planned Completion

Module Name Status Date

Real-Time Accuracy not optimal, needs


2/03/2025
Prediction improvement.

Gesture-to-text conversion needs


NLP Integration 18/02/2025
refinement.

Convert recognized gestures into 18/02/2025


Voice Output
speech.
IEEE Paper
● The IEEE paper is not yet completed as real-time detection is still being improved.
● The progress so far covers data collection, model training, and evaluation.

Completed Work
● Data Collection: Used MediaPipe to extract and normalize hand and face landmarks.
● Model Training: Implemented an LSTM + CNN model for gesture recognition.
● Evaluation: Achieved good accuracy in testing, analyzed misclassifications using a confusion matrix.

Pending Work
● Real-Time Detection: Accuracy is not yet optimal, with issues in prediction speed and misclassification.
● Finalizing IEEE Paper: Need to refine real-time results before completing the paper.

Next Steps
● Optimize real-time detection for better accuracy.
● Expand dataset and refine gesture recognition.
● Improve model efficiency to reduce lag.
THANK YOU!

You might also like