The solution enables highly accurate human detection on the Edge by running powerful ML on low-cost and energy-efficient hardware. Plumerai's software is capable of running on virtually any camera device as a software update, saving expensive edge computing chips and cloud costs. With a U.S. patent application for "neural network systems and methods" and an award-winning inference engine tool, Plumerai is an innovation leader in the industry.

Plumerai is an innovative provider of AI software solutions for human detection on cameras. The challenge with human detection on cameras is that it requires either a performant AI chip on each camera for edge computing or a connection to an AI on the cloud - both of which have significant drawbacks. Deploying performant chips on each camera is expensive (from $30 per camera), requires a lot of power and thus the need for power cables, and the increased heat dissipation requires larger cameras. Deploying AI in the cloud is also expensive (~$20 per camera per year), requires a lot of power (~300 mW) and results in additional latency of ~1 second and upload of ~2 Mbps per camera, which slows down Wi-Fi. In addition, data protection laws impose restrictions on video streaming in the cloud.

Plumerai's solution enables high-precision human detection through AI at the edge. The solution runs on powerful ML software on low-cost and energy-efficient hardware. The AI models can run on microcontrollers based on Arm Cortex-M CPUs, where the processor load is minimal and therefore computing resources are available for additional applications on the device. Plumerai collects and labels its own data sets, while the data pipeline continuously improves the reliability and accuracy of the models. The model is trained with over 30 million images and recognizes people with very high accuracy under different conditions (indoor/outdoor, dim/bright light, moving objects, difficult backgrounds).