analiza aiqbench
This research addresses the critical challenge of deploying large vision-language models on resource-constrained embedded fire sensors by utilizing knowledge distillation to compress model size without sacrificing safety-critical reasoning. The study demonstrates that specific lightweight architectures, such as Qwen2.5-0.5B, can effectively balance latency and memory usage while maintaining high accuracy in semantic fire detection. These findings offer essential guidance for integrating advanced AI into real-world safety systems where edge deployment is mandatory.
Distilling Vision-Language Models for On-Device Fire Understanding
arXiv:2609.05782v1 Announce Type: new Abstract: Vision-language models (VLMs) offer a promising alternative to conventional fire detection systems by reasoning about the semantic context of a scene and thus reducing false alarms, yet their large model size makes deployment on embedded fire sensors impractical. In this paper, we study how domain-specialized VLMs can be compressed for fully on-device deployment without losing the safety-critical behavior required for fire detection. We develop a teacher-student knowledge distillation framework in which large VLMs fine-tuned for fire understanding can be distilled into lightweight students. Experiments across multiple VLM families and model scales show that compact students preserve most of their teachers' fire-understanding capability. We further deploy the distilled models on our commercial Detectium fire detection sensor and jointly evaluate reasoning accuracy, latency, and memory usage. The results show that compression and deployment affect not only accuracy but also model failure modes, with Qwen2.5-0.5B providing the strongest overall deployment trade-off. Our findings provide broader guidance for deploying domain-specialized VLMs in resource-constrained, safety-critical settings.
Abstract:Vision-language models (VLMs) offer a promising alternative to conventional fire detection systems by reasoning about the semantic context of a scene and thus reducing false alarms, yet their large model size makes deployment on embedded fire sensors impractical. In this paper, we study how domain-specialized VLMs can be compressed for fully on-device deployment without losing the safety-critical behavior required for fire detection. We develop a teacher-student knowledge distillation framework in which large VLMs fine-tuned for fire understanding can be distilled into lightweight students. Experiments across multiple VLM families and model scales show that compact students preserve most of their teachers' fire-understanding capability. We further deploy the distilled models on our commercial Detectium fire detection sensor and jointly evaluate reasoning accuracy, latency, and memory usage. The results show that compression and deployment affect not only accuracy but also model failure modes, with Qwen2.5-0.5B providing the strongest overall deployment trade-off. Our findings provide broader guidance for deploying domain-specialized VLMs in resource-constrained, safety-critical settings.
Artificial Intelligence (cs.AI)
arXiv:2609.05782 [cs.AI]
(or arXiv:2609.05782v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.05782 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 255 cuvinte
Subiecte conexe
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · 13 oct. 2026
[AINews] not much happened today
Latent Space · 10 sept. 2026
Align-KD: Distilling Cross-Modal Alignment Knowledge for Mobile Vision-Language Model Enhancement
arXiv · 3 iun. 2026
Physical AI’s moment has arrived – but moving from demo to deployment is the hard part. AWS wants to fix that
ZK Research · 24 aug. 2026