HumP-KD: A Hybrid Uncertainty-Aware Multi-Stage Progressive Knowledge Distillation Framework for Efficient Fire Classification
arXiv:2606.14684v1 Announce Type: cross Abstract: Real-time fire classification systems require models that are simultaneously accurate, computationally efficient, and deployable on resource-constrained hardware. This work proposes \textbf{HumP-KD}, a Hybrid Uncertainty-aware Multi-stage Progressive Knowledge Distillation framework for efficient fire classification. Two datasets, FlameVision and Dataset-II, containing 8,600 and 31,309 images, are used. Various CNN and transformer baselines are applied under standard preprocessing, online augmentation, Gaussian noise and motion blur robustness
The continuous drive for real-time AI solutions, especially in safety-critical applications, necessitates efficient model deployment on resource-constrained hardware.
This work represents progress in deploying advanced AI capabilities in edge environments, crucial for applications like autonomous safety systems and remote monitoring.
The development of more efficient deep learning models will accelerate the deployment of AI in resource-limited settings, expanding the practical reach of AI vision systems.
- · Edge AI hardware manufacturers
- · Emergency services technlogy providers
- · Computer vision developers
- · Safety and surveillance sectors
- · Companies relying solely on high-compute cloud AI for real-time applications
More widespread and reliable real-time fire detection systems become feasible.
The cost of deploying AI-powered monitoring solutions decreases, leading to broader adoption across various industries.
Increased public and industrial safety can result from ubiquitous, efficient real-time threat detection capabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG