Towards Blind and Low-Vision Accessibility of Lightweight VLMs and Custom LLM-Evals
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912707344596992 |
|---|---|
| author | Baghel, Shruti Singh Rathore, Yash Pratap Singh Jena, Sushovan Pradhan, Anurag Shukla, Amit Bhavsar, Arnav Goyal, Pawan |
| author_facet | Baghel, Shruti Singh Rathore, Yash Pratap Singh Jena, Sushovan Pradhan, Anurag Shukla, Amit Bhavsar, Arnav Goyal, Pawan |
| contents | Large Vision-Language Models (VLMs) excel at understanding and generating video descriptions but their high memory, computation, and deployment demands hinder practical use particularly for blind and low-vision (BLV) users who depend on detailed, context-aware descriptions. To study the effect of model size on accessibility-focused description quality, we evaluate SmolVLM2 variants with 500M and 2.2B parameters across two diverse datasets: AVCaps (outdoor), and Charades (indoor). In this work, we introduce two novel evaluation frameworks specifically designed for BLV accessibility assessment: the Multi-Context BLV Framework evaluating spatial orientation, social interaction, action events, and ambience contexts; and the Navigational Assistance Framework focusing on mobility-critical information. Additionally, we conduct a systematic evaluation of four different prompt design strategies and deploy both models on a smartphone, evaluating FP32 and INT8 precision variants to assess real-world performance constraints on resource-limited mobile devices. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_10615 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards Blind and Low-Vision Accessibility of Lightweight VLMs and Custom LLM-Evals Baghel, Shruti Singh Rathore, Yash Pratap Singh Jena, Sushovan Pradhan, Anurag Shukla, Amit Bhavsar, Arnav Goyal, Pawan Computer Vision and Pattern Recognition Computation and Language I.2.7 Large Vision-Language Models (VLMs) excel at understanding and generating video descriptions but their high memory, computation, and deployment demands hinder practical use particularly for blind and low-vision (BLV) users who depend on detailed, context-aware descriptions. To study the effect of model size on accessibility-focused description quality, we evaluate SmolVLM2 variants with 500M and 2.2B parameters across two diverse datasets: AVCaps (outdoor), and Charades (indoor). In this work, we introduce two novel evaluation frameworks specifically designed for BLV accessibility assessment: the Multi-Context BLV Framework evaluating spatial orientation, social interaction, action events, and ambience contexts; and the Navigational Assistance Framework focusing on mobility-critical information. Additionally, we conduct a systematic evaluation of four different prompt design strategies and deploy both models on a smartphone, evaluating FP32 and INT8 precision variants to assess real-world performance constraints on resource-limited mobile devices. |
| title | Towards Blind and Low-Vision Accessibility of Lightweight VLMs and Custom LLM-Evals |
| topic | Computer Vision and Pattern Recognition Computation and Language I.2.7 |
| url | https://arxiv.org/abs/2511.10615 |