Recently, a pioneering research team from Hubei University’s School of Integrated Circuits—comprising Professor Wang Hao, Associate Professor Han Wei, Professor Shen Liangping, Professor Ma Guokun, and Dr. Wan Houzhao—has achieved a significant breakthrough in in-sensor computing for intelligent flame monitoring. Their study, titled “Ferroelectric Optoelectronic Sensor for Intelligent Flame Detection and In-Sensor Motion Perception,” was published in Nano-Micro Letters, a leading international journal in nanotechnology and microelectronics.

Addressing the limitations of conventional flame detection systems—particularly their poor performance in low-light conditions and dynamic recognition—the team developed a ferroelectric optoelectronic sensor array (Fe-OES) based on a Ga₂O₃/In₂Se₃ heterojunction. Leveraging the ferroelectric polarization effect, this innovation dramatically enhances solar-blind ultraviolet (200–280 nm) detection sensitivity. Under 255 nm ultraviolet illumination, the 5×5-pixel sensor array demonstrated a detectivity of 4.91×10¹⁷ Jones, outperforming traditional gallium oxide-based devices by over two orders of magnitude.

The research achieved three key advancements:
1. All-Weather Flame Monitoring System: By integrating terminal-cloud interaction technology, the system achieved a 100% alarm success rate within 25 seconds across diverse outdoor scenarios.
2. Flame Motion Trajectory Recognition: Utilizing a lightweight convolutional neural network (CNN), the team attained 96.47% accuracy in tracking flame movement.
3. Photosensitive Artificial Nervous System: Inspired by biological neural mechanisms, this system enables early detection of weak flame signals with 90.51% accuracy, mimicking human sensory perception.

This technology offers a transformative approach to fire prevention in forests, industrial facilities, and other high-risk environments. The findings have been secured through multiple national invention patents, with practical applications already realized in fire early-warning inspection robots.