RADMANTIS PRESS KIT — CAPTIONS & CREDITS

All assets are cleared for editorial use in connection with this announcement, with credit as noted. Please do not alter or crop the AI-annotated imagery in ways that misrepresent the technology. For high-resolution originals or additional footage, contact Dr. Moira van Staaden, mvs@radmantis.com.

Logos

Radmantis/Radmantis_Logo.png / .svg
Radmantis Inc. logo. Credit: Radmantis Inc.

Verdantas/Verdantas-rgb-color-dark.png (included with permission)
Verdantas logo. Used with permission. Credit: Verdantas Flow Labs.

Headshots

van Staaden-M_Radmantis_CEO.jpg
Dr. Moira van STaaden, CEO & Cofounder, Radmantis Inc. Credit: Radmantis Inc.

Huber-Robert_Radmantis_CTO.jpg
Dr. Robert Huber, CTO & Cofounder, Radmantis Inc. Credit: Radmantis Inc.

Imagery

FlowLabs_Inference.png
Underwater scene with Radmantis AI inference applied — individual fish detected, classified and tracked. Credit: Radmantis Inc.

FlowLabs_View.png
Underwater scene through Viewing Window. Credit: Radmantis Inc.

FlowLabs_Setup.jpg
The Radmantis validation study underway at Verdantas Flow Labs' Alden Campus hydraulics laboratory in Holden, Massachusetts. The experimental flume channels water through a controlled fish-passage model, with mesh screens and framed enclosures guiding trout through the imaging zone where the Radmantis underwater vision system captures them. A laptop workstation at the flume's edge runs the platform's analysis during trials. Credit: Radmantis Inc. / Verdantas Flow Labs.

Video

Inferencing_notracks-a_H265.mp4
Fish detection and trout classification by the Radmantis underwater vision system during the validation study — individual trout identified and classified as they move through the field of view. (30 seconds; H.265 encoded.) Credit: Radmantis Inc.

Inferencing_tracks-b_H265.mp4
Radmantis model inference in a dense, visually complex scene — individual trout are resolved within a tightly clustered group and tracked over time, each fish's movement traced with a visual path indicator. (60 seconds; H.265 encoded.) Credit: Radmantis Inc.

thumbnails/Inferencing_notracks-a_H265_thumb.png
Still frame from Radmantis species-tracking footage in Inferencing_notracks-a_H265.mp4. Credit: Radmantis Inc.

thumbnails/Inferencing_tracks-b_H265_thumb.png
Still frame from Radmantis species-tracking footage in Inferencing_tracks-b_H265.mp4. Credit: Radmantis Inc.

Study Fast Facts (for reference in captions or copy)
99% accuracy reconstructing fish species composition (exact in 36 of 40 single-blind trials)
95.5% fish detection reliability (mAP50) under dense, high-movement conditions
Conducted at Verdantas Flow Labs' Alden Campus hydraulics laboratory, Holden, Massachusetts
Supported by the U.S. Department of Energy's Water Power Technologies Office via ENERGYWERX and the Hydropower Testing Network (HyTN)
