Gloss-free SLT via LLM/VLM

1. 4. Datasets for Gloss-Free SLT

1.1. 4.1. Arabic Sign Language Datasets

1.2. 4.2. Other Regional Datasets

1.2.1. 4.2.1. Indo-Pak Sign Language (ISL) Datasets

1.2.2. 4.2.2. Indonesian Sign Language (SIBI) Datasets

1.2.3. 4.2.3. African Sign Languages

1.2.4. 4.2.4. Underrepresented Gloss-free SLT Datasets

1.3. 4.3. Dataset Requirements for Gloss-Free SLT

2. 5. Benchmark Performance Comparison in Chronological manner

2.1. 5.1. PHOENIX14T Dataset

2.2. 5.2 CSL-Daily Dataset

2.3. 5.3 How2sign Dataset

3. 7. Real-Time Applicability and Bidirectional Communication in Gloss-Free SLT

4. 8. Current Landscape of Sign Language Applications in Saudi Arabia

4.1. 8.1. Survey Findings on DHH Application Usage

4.2. 8.2. Efhamni’s Approach and Limitations

5. 1. Introduction

6. 2. Transition from Gloss-based to Gloss-free SLT

6.1. 2.1. Gloss-based SLT

6.2. 2.2. Gloss-free SLT

6.3. 2.3. Recent Advances and the Road Ahead

7. 6. Evaluation Metrics for SLT

7.1. 6.1 Evaluation Metrics for SLT

7.2. Human Centric Evaluation

8. Conclusion

9. 3. Gloss-Free SLT models using LLM/VLM

9.1. 3.1. GFSLT-VLP

9.2. 3.2. FLA-LLM

9.3. 3.3. SignLLM

9.4. 3.4. Sign2GPT

9.5. 3.5. SignCL

9.6. 3.6. SpaMo

9.7. 3.7. MMSLT

9.8. 3.8. DiffSLT

9.9. 3.9. LLaVA-SLT