The demand for privacy-first camera
As cities and homes become increasingly intelligent and connected, the demand for image-based data collection is rising rapidly, along with heightened concerns about privacy. Calumino has placed privacy at the forefront by developing the world’s first affordable thermal sensor specifically engineered for mass-market camera modules. Calumino’s sensor delivers powerful, real-time insights while inherently safeguarding individual privacy. To restore public trust in data-driven innovation without compromising performance, the Australian company partnered with ST to leverage its state-of-the-art edge AI technology.
Challenge
- Develop sensors that provide meaningful data, such as occupancy and activity detection, without capturing identifiable images or transmitting data to the cloud
- Enable real-time data analysis directly on the camera itself while maintaining low power consumption to preserve battery life
Solution
- STM32N6 with ST’s in-house developed neural processing unit to enable privacy-first thermal sensors
- ST’s free software tools to convert and embed Calumino’s proprietary AI models
Impact
- Alleviate citizen concerns about surveillance and misuse of data
- Enable wider acceptance and adoption of smart devices
How on-device AI enables data privacy
Traditional cameras capture detailed images that can easily reveal personal identities. Additionally, cloud-based processing often requires transmitting this sensitive data to remote data centers operated by third parties. In response to growing public concerns about privacy, Calumino applied its core expertise in thermal sensing, which detects only heat signatures—and represents individuals as color-coded shapes rather than identifiable faces. However, processing even these anonymized thermal images in the cloud could still pose privacy risks. The solution was clear: all data processing needed to happen directly on the device. This requirement is precisely where ST, a trusted leader in edge AI innovation, became an essential partner.
“
The STM32N6 has been critical to delivering high-performance AI at the edge while maintaining low power consumption and privacy.”
Yalcin Bulut, CEO, Calumino
Why Calumino settled on the STM32N6
The STM32N6 microcontroller delivers efficient, secure on-device AI processing. It offers the performance Calumino needs to enable real-time edge AI, combined with low power consumption to support always-on devices. The microcontroller’s low power capabilities partners well with the ultra-low power Calumino technology (based on passive sensing principles), keeping the overall system power consumption low.
And there’s more. This advanced microcontroller equipped with an ST in-house developed neural processing unit is affordable and scalable for mass deployment.
On the development side, Calumino was able to run its own algorithms directly on the sensor, ensuring end-user privacy is fully protected. The ST Edge AI Suite further streamlined the process, allowing Calumino to convert, optimize, and deploy its models through the STM32Cube.AI, benchmarking and code-generation tools. This ensured predictable, high-quality model performance on the microcontroller while significantly reducing engineering overhead.
“
STM32N6 gives us a stable, reliable foundation as we scale from prototypes to millions of units.”
Dawid Oosthuizen, Head of Software, Calumino
Who thought heat could do all that?
Calumino’s privacy-first thermal sensors are making waves across diverse industries. They help optimize energy use by monitoring occupancy in smart buildings. In senior care centers they track falls, and daily activity patterns without intrusive cameras, preserving residents’ dignity. They even keep a close eye on assets and detect anomalies in manufacturing. By combining robust privacy with powerful insights, Calumino is redefining how smart technology can respect and uphold individual privacy.
