News: iPod Inventor and Nest Founder Tony Fadell Backs Plumerai — Press release .

World's most accurate AI for Renesas RA and RZ.

Plumerai provides fully optimized AI solutions for Familiar face identification and People detection. Our models are accurate, fast, and tiny. Trained on a dataset of 30 million images and videos, the AI offers precise detections in any scenario. The Renesas RA8 Arm Cortex-M85 MCU runs People Detection at 13 fps, and Familiar Face Identification at 4 fps. Renesas RZ MPUs are fully supported as well.

RA8 ARM CORTEX-M85 PEOPLE DETECTION

13 frames⁄s

WITH A TINY FOOTPRINT

1.5 MB

People Detection

  • Detects each person in view, even if partially occluded.
  • Tracks people and assigns up to 20 unique IDs.
  • Indoor, outdoor, NIR lighting.
  • Detection distance of more than 20m / 65ft.
  • Trained with 30 million labeled images and videos.
  • Extensively validated on diverse people and settings.
  • Supports lenses up to 180 FOV.

Familiar Face Identification

  • Identifies individuals and strangers, even from a distance.
  • Offers both automatic and manual enrollment options.
  • Industry-leading accuracy: very few missed or incorrect identifications.
  • Detects strangers to reliably enable appropriate security responses.
  • Supports indoor, outdoor, and cameras with night vision
  • Detection distance exceeding 10m / 30ft.
  • Trained with 30 million labeled images and videos.
  • Validated across diverse demographics and environments.
  • Supports lenses up to 180 FOV.

Optimized for Renesas RA8 and RZ.

People Detection on RA8

  • Arm Cortex-M85 with Helium
  • Latency: 73.3 ms (single core)
  • Peak RAM usage: 289 KiB
  • Binary size: 1.2 MiB total

Familiar Face Identification on RA8

  • Arm Cortex-M85 with Helium
  • Latency: 255 ms (single core)
  • Peak RAM usage: 617 KiB
  • Binary size: 5.5 MiB total

Familiar Face Identification on RZ

  • Single Arm Cortex-A55 with Neon
  • Latency: 54 ms (single core)
  • Binary size: 5.5 MiB total

Get access to the Plumerai demos and libraries.

Integration into your system is easy with our C++ and Python APIs.

Thank you for submitting the form. We’ll get in touch as soon as possible.

There was an error submitting the form. Please try again later.

There are many more applications where detecting or identifying people provides value. We're helping robots to detect humans, PIR-based motion sensors to become more accurate, detecting pedestrians to increase the efficiency of traffic lights and elevators, and many more.