SHENZHEN, China, Sept. 17, 2026 /PRNewswire/ -- As AI expands beyond content generation to perceiving and understanding the physical world, intelligent sensing is becoming increasingly important across industrial applications. At CIOE 2026, Raytron, a global leader in infrared thermal imaging, unveiled its Infrared Vision-Language Model, extending infrared imaging beyond image acquisition toward intelligent understanding. Raytron also introduced Falcon 500, its third-generation AI image processing chip, and showcased the latest advancements in its 8 µm and 12 µm infrared detector portfolios.
Why Does Infrared Imaging Need a Dedicated Vision-Language Model?
Unlike visible-light imaging, which primarily captures reflected light, infrared thermal imaging detects radiation emitted by objects, revealing unique information such as temperature, heat distribution, and thermal signatures. Most general-purpose vision models are trained primarily on large-scale visible-light data and are not specifically designed to interpret the unique characteristics of infrared imagery. Raytron's Infrared Vision-Language Model is designed to address this challenge through:
- 5M+ Proprietary Infrared Data: More than five million high-quality multimodal infrared data samples cover specialized scenarios including industrial temperature measurement, night vision, and gas detection.
- Dedicated Infrared Visual Encoder: Designed specifically for infrared imaging, the dedicated encoder enhances the model's ability to capture and interpret thermal distributions and infrared signal features.
- Industry-Specific Training: The model is trained for application-specific scenarios, such as gas detection, electrical inspection and perimeter monitoring. Training data covers gases such as methane, sulfur dioxide, ethanol, R-134a, R-152a, and sulfur hexafluoride, while accounting for gas plume patterns, dispersion behavior, and background interference to improve the identification of potential leakage areas.
- Edge AI Deployment: Enables local deployment and hardware acceleration of AI models on edge chips, allowing devices to perform real-time intelligent sensing without relying on cloud connectivity. This reduces latency, enhances data security, and helps optimize overall hardware costs.
Advancing Infrared Sensing with Falcon 500 and 8µm HD Infrared Detectors
The launch was accompanied by further advances in infrared image processing and detectors. The company unveiled Falcon 500, its third-generation infrared AI image processing chip, bringing enhanced image quality, intelligent sensing, multimodal fusion, and precision temperature measurement.
At the exhibition, Raytron also showcased its expanded 8 µm lineup, spanning resolutions from 640×512 to 1280×1024, alongside upgraded 12 µm products offering NETD as low as 15 mK.
For Further Information
Email: sales@raytrontek.com
Website: https://en.raytrontek.com
LinkedIn: Raytron Technology Co., Ltd.
View original content to download multimedia:https://www.prnewswire.co.uk/news-releases/raytron-unveils-infrared-vision-language-model-at-cioe-2026-advancing-thermal-imaging-from-perception-to-understanding-302881706.html
