{"slug": "small-yet-assistive-spatially-aware-post-training-for-low-vision", "title": "Small yet Assistive: Spatially-Aware Post-Training for Low Vision", "summary": "Researchers released Smol-VL-BLV, a 500M-parameter vision-language model for blind and low-vision users that combines teacher-student distillation with Group Relative Policy Optimization (GRPO) using a composite BLV reward for directional language, metric distances, and hazard detection. The model improves the Spatial score by 19.3% and the Social score by 14.8% over baseline, raises OCR-Bench by 101.5% and TextVQA accuracy by 44.2%, and runs entirely on-device at approximately 450 MB on a mid-range Android smartphone via mixed-precision quantization. The model, dataset, and code are publicly released at https://smol-vl-blv.github.io/Smol-VL-BLV-website/.", "body_md": "arXiv:2609.28757v1 Announce Type: new \nAbstract: An estimated 1 billion people worldwide live with vision impairment, yet current vision-language models (VLMs) produce descriptions too vague for safe navigation by blind and low-vision (BLV) users. Large VLMs can generate high-quality audio-description-compliant narrations but cannot run on mobile devices; small VLMs offer competitive latency but lack spatial detail, directional cues, and hazard awareness for navigational assistance. We present Smol-VL-BLV, a compact VLM for blind and low-vision users that closes this gap using a 500M decoder transformer model and two post-training mechanisms: (1) teacher-student distillation and (2) Group Relative Policy Optimization (GRPO) with a composite BLV reward targeting directional language, metric distances, and hazard detection. Because multi-stage post-training can induce catastrophic forgetting, we add a lightweight finetuning stage after the last stage GRPO finetuning to recover general descriptive quality while preserving BLV-specific spatial grounding. Our best model substantially outperforms the baseline across various benchmarks, including tasks: VQA, BLV captioning, OCR, and latency. Compared with the baseline for relative improvement, it improves the Spatial score gain of 19.3%, and the Social score gain of 14.8%. It also increases OCR-Bench by 101.5%, and raises TextVQA accuracy by 44.2%. These results show that BLV-focused post-training improves both accessibility-specific spatial grounding and general visual-text reasoning. Deployed on a mid-range Android smartphone via Mixed-Precision Quantization, the model remains approx. 450 MB and runs entirely on-device, offline and without network dependency, generating descriptions with latency dependent on host hardware capabilities. Our model, dataset, and code is publicly released at https://smol-vl-blv.github.io/Smol-VL-BLV-website/", "url": "https://wpnews.pro/news/small-yet-assistive-spatially-aware-post-training-for-low-vision", "canonical_source": "https://arxiv.org/abs/2609.28757", "published_at": "2026-09-25 04:00:00+00:00", "updated_at": "2026-09-25 04:01:29.082814+00:00", "lang": "en", "topics": ["computer-vision", "natural-language-processing", "ai-research", "ai-products", "machine-learning"], "entities": ["Smol-VL-BLV", "Group Relative Policy Optimization", "arXiv", "Android"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/small-yet-assistive-spatially-aware-post-training-for-low-vision", "markdown": "https://wpnews.pro/news/small-yet-assistive-spatially-aware-post-training-for-low-vision.md", "text": "https://wpnews.pro/news/small-yet-assistive-spatially-aware-post-training-for-low-vision.txt", "jsonld": "https://wpnews.pro/news/small-yet-assistive-spatially-aware-post-training-for-low-vision.jsonld"}}