Table 1 Summary of existing studies on SLR systems for assisting hearing and speech-impaired individuals.

Rethick et al¹¹.

To develop a real-time hand gesture recognition system to empower the deaf and mute community.

CNN, Real-time gesture detection, Custom CNN architecture, Fine-tuning for accuracy

Diverse ASL gesture images (29 classes, 87,000 images)

Accuracy of 99.11%, Real-Time Performance

Assiri and Selim¹²

To develop a robust hand gesture recognition system for helping hearing-impaired individuals.

ABF, ST, CNN-BiLSTM, SBOA

Traffic Police Gesture Dataset

Accuracy of 99.25%, Performance Validation

Kumar, Reddy, and Swetha¹³

To develop a real-time system for converting gestures into text and speech to enhance communication for the Deaf and Hard of Hearing community.

CNN, RNN

Hindi SL Gesture Dataset

Accuracy, Real-Time Performance

Harshini et al¹⁴.

To develop an SLR system for SL users in digital environments.

RF, Comparison with CNN and KNN, Integration with Conversational AI

Diverse SL Gesture Dataset

Accuracy of 0.9961, User Responsiveness

Allehaibi¹⁵

To develop a robust SLR system using optimized DL models for accurate gesture classification.

RSLR-CEWODL, ResNet-101, WOA, DBN

SL Gesture Datasets

Accuracy, Performance Evaluation

Kumar et al¹⁶.

To enhance ISL recognition by integrating DL with manually designed features for improved accuracy and robustness.

Deep CNN, DL

Extensive ISL Dataset

Recognition Accuracy, Robustness

Hariharan et al¹⁷.

To develop an accurate ASL recognition system.

MCNN, ResNet-101

ASL Hand Gesture Images (36 Signs)

Accuracy (97%), False Positive Rate (0.05%)

Almjally and Almukadi¹⁸

To develop an optimized DL technique for accurate and automatic SLR.

BF, ResNet-152, Bi-LSTM, HHO

SL Dataset

Precision, Recall, Accuracy, and F1-Score of 94.72%, 94.74%, 98.95, and 94.72%

Kaur et al¹⁹.

To develop a real-time SL-to-speech conversion system.

InceptionResNetV2 DL, Hand keypoint extraction, Python image processing, Model training with epochs

7200 Images, 24 Alphabet Classes (Excluding ‘J’ and ‘Z’)

Training And Validation Accuracy

Almjally et al²⁰.

To enhance accurate and real-time SLR using an attention-driven hybrid DL technique with feature fusion.

CLAHE, CED, ST, ConvNeXt-Large, ResNet50, CNN-BiLSTM-A

SL Dataset

Accuracy of 98.10%, Precision of 95.28%, Recall of 95.28%, and F1-Score of 95.28%

Jagdish and Raju²¹

To develop a DL-based system for accurate detection and recognition of SL gestures.

CNN Model Training, Sign-to-Text Conversion, Voice Output Integration

SL Gesture Images

Accuracy, Accessibility

Maashi, Iskandar, and Rizwanullah²²

To develop an intelligent SLR system using a hybrid DL method to assist hearing-impaired individuals.

SACHI, MobileNetV3, CNN-BiGRU-A, AROA

ISL Dataset

Precision of 91.54%, Recall of 93.21%, Accuracy of 99.19%, and F-Score of 91.87%, respectively.

Ilakkia et al²³.

To develop a real-time ISL recognition system to translate ISL gestures into text for the deaf community.

DL, ResNet-50

Unique ISL Dataset

Accuracy, Real-Time Performance

Mosleh et al²⁴.

To develop a robust, real-time, bidirectional ArSL translation system to enhance communication for deaf individuals.

ArSL, CNN

ArSL Dataset and Arabic Data Dictionary

Accuracy, Processing Efficiency

Thakkar, Kittur, and Munshi²⁵

To develop a multilingual SL translation system to facilitate communication between hearing, visually impaired, and auditory-impaired individuals.

YOLOv5, LSTM-GRU, RF, OpenCV & MediaPipe integration, Auto-tokenizer and Adam optimizer

Multilingual SL Images

Translation Accuracy, Processing Speed

Dhaarini, Sanjai, and Sandosh²⁶

To develop a real-time SL Detection and Assistive System for deaf and mute individuals.

SLDAS, CV, YOLOv10

ASL Gesture Dataset

Accuracy, Real-Time Performance

Choudhari et al²⁷.

To develop a platform-independent web-based system for real-time ISL translation into text.

CNN, Leaky ReLU, Adam Optimiser

1200 Images, 35 Classes (26 Alphabets + 9 Numbers)

Accuracy of 97%, Real-time Performance