Lightweight Operations for Visual Speech Recognition
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929702286917632 |
|---|---|
| author | Panagos, Iason Ioannis Sfikas, Giorgos Nikou, Christophoros |
| author_facet | Panagos, Iason Ioannis Sfikas, Giorgos Nikou, Christophoros |
| contents | Visual speech recognition (VSR), which decodes spoken words from video data, offers significant benefits, particularly when audio is unavailable. However, the high dimensionality of video data leads to prohibitive computational costs that demand powerful hardware, limiting VSR deployment on resource-constrained devices. This work addresses this limitation by developing lightweight VSR architectures. Leveraging efficient operation design paradigms, we create compact yet powerful models with reduced resource requirements and minimal accuracy loss. We train and evaluate our models on a large-scale public dataset for recognition of words from video sequences, demonstrating their effectiveness for practical applications. We also conduct an extensive array of ablative experiments to thoroughly analyze the size and complexity of each model. Code and trained models will be made publicly available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_04834 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Lightweight Operations for Visual Speech Recognition Panagos, Iason Ioannis Sfikas, Giorgos Nikou, Christophoros Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language Machine Learning Visual speech recognition (VSR), which decodes spoken words from video data, offers significant benefits, particularly when audio is unavailable. However, the high dimensionality of video data leads to prohibitive computational costs that demand powerful hardware, limiting VSR deployment on resource-constrained devices. This work addresses this limitation by developing lightweight VSR architectures. Leveraging efficient operation design paradigms, we create compact yet powerful models with reduced resource requirements and minimal accuracy loss. We train and evaluate our models on a large-scale public dataset for recognition of words from video sequences, demonstrating their effectiveness for practical applications. We also conduct an extensive array of ablative experiments to thoroughly analyze the size and complexity of each model. Code and trained models will be made publicly available. |
| title | Lightweight Operations for Visual Speech Recognition |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2502.04834 |