Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition
Abstract Sign language recognition (SLR) systems continue to face significant challenges in accurately interpreting dynamic gestures, particularly for underrepresented languages like Kannada sign language (KSL). This study presents a novel hybrid deep learning architecture that synergistically combi...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | International Journal of Computational Intelligence Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44196-025-00922-4 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849331768666095616 |
|---|---|
| author | Gurusiddappa Hugar Ramesh M. Kagalkar Abhijit Das |
| author_facet | Gurusiddappa Hugar Ramesh M. Kagalkar Abhijit Das |
| author_sort | Gurusiddappa Hugar |
| collection | DOAJ |
| description | Abstract Sign language recognition (SLR) systems continue to face significant challenges in accurately interpreting dynamic gestures, particularly for underrepresented languages like Kannada sign language (KSL). This study presents a novel hybrid deep learning architecture that synergistically combines convolutional neural networks (CNNs), hand keypoints (HKPs), long short-term memory (LSTM) networks, and transformers to achieve robust spatial-temporal-contextual learning for KSL recognition. Developed on a newly curated dataset of 1080 medical-domain KSL gestures, our model addresses critical gaps in dataset diversity and model generalizability. The proposed framework demonstrates superior performance with 97.6% training accuracy, 96.75% validation accuracy, and 81% testing accuracy on unseen data—outperforming conventional CNN-LSTM (46%) and HKP-LSTM (71%) baselines. By hierarchically integrating CNN-extracted spatial features, HKP-derived structural priors, LSTM-processed temporal dynamics, and Transformer-modeled long-range dependencies, this work establishes a new benchmark for KSL recognition while providing a scalable solution for real-world healthcare and assistive technology applications. |
| format | Article |
| id | doaj-art-cc7414da711f4b16bbdab043f5df8305 |
| institution | Kabale University |
| issn | 1875-6883 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Springer |
| record_format | Article |
| series | International Journal of Computational Intelligence Systems |
| spelling | doaj-art-cc7414da711f4b16bbdab043f5df83052025-08-20T03:46:24ZengSpringerInternational Journal of Computational Intelligence Systems1875-68832025-07-0118112310.1007/s44196-025-00922-4Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language RecognitionGurusiddappa Hugar0Ramesh M. Kagalkar1Abhijit Das2Computer Science and Engineering, AGMR College of Engineering and Technology Varur, (Visvesvaraya Technological University Belagavi)Information Science and Engineering, Nagarjuna College of Engineering and Technology Bengaluru, (Visvesvaraya Technological University Belagavi)Department of Information Technology, Manipal Institute of Technology Bengaluru, Manipal Academy of Higher EducationAbstract Sign language recognition (SLR) systems continue to face significant challenges in accurately interpreting dynamic gestures, particularly for underrepresented languages like Kannada sign language (KSL). This study presents a novel hybrid deep learning architecture that synergistically combines convolutional neural networks (CNNs), hand keypoints (HKPs), long short-term memory (LSTM) networks, and transformers to achieve robust spatial-temporal-contextual learning for KSL recognition. Developed on a newly curated dataset of 1080 medical-domain KSL gestures, our model addresses critical gaps in dataset diversity and model generalizability. The proposed framework demonstrates superior performance with 97.6% training accuracy, 96.75% validation accuracy, and 81% testing accuracy on unseen data—outperforming conventional CNN-LSTM (46%) and HKP-LSTM (71%) baselines. By hierarchically integrating CNN-extracted spatial features, HKP-derived structural priors, LSTM-processed temporal dynamics, and Transformer-modeled long-range dependencies, this work establishes a new benchmark for KSL recognition while providing a scalable solution for real-world healthcare and assistive technology applications.https://doi.org/10.1007/s44196-025-00922-4Sign language recognition (SLR)Kannada sign language (KSL)Hybrid deep learningConvolutional neural networks (CNN)Hand keypoints (HKP)Long short term memory (LSTM) |
| spellingShingle | Gurusiddappa Hugar Ramesh M. Kagalkar Abhijit Das Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition International Journal of Computational Intelligence Systems Sign language recognition (SLR) Kannada sign language (KSL) Hybrid deep learning Convolutional neural networks (CNN) Hand keypoints (HKP) Long short term memory (LSTM) |
| title | Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition |
| title_full | Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition |
| title_fullStr | Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition |
| title_full_unstemmed | Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition |
| title_short | Comparative Study of Hybrid Deep Learning Models for Kannada Sign Language Recognition |
| title_sort | comparative study of hybrid deep learning models for kannada sign language recognition |
| topic | Sign language recognition (SLR) Kannada sign language (KSL) Hybrid deep learning Convolutional neural networks (CNN) Hand keypoints (HKP) Long short term memory (LSTM) |
| url | https://doi.org/10.1007/s44196-025-00922-4 |
| work_keys_str_mv | AT gurusiddappahugar comparativestudyofhybriddeeplearningmodelsforkannadasignlanguagerecognition AT rameshmkagalkar comparativestudyofhybriddeeplearningmodelsforkannadasignlanguagerecognition AT abhijitdas comparativestudyofhybriddeeplearningmodelsforkannadasignlanguagerecognition |