YouTuber Builds Edge AI Microscope with Jetson

Hackster.io reports that YouTuber That Project built a custom digital microscope around a DFRobot HuskyLens 2 AI camera and an NVIDIA Jetson Orin Nano, replacing the HuskyLens 2's stock lens with a long microscope lens for roughly 30x magnification, mounted in a custom 3D-printed enclosure. Two separate USB connections were needed, one for control commands and one for the video stream, to work around issues with the Jetson's serial interface. Captured images are passed to a locally running Gemma vision model, which can analyze specimens entirely offline. In an initial test on pollen placed on transparent tape over black paper for contrast, the model offered general observations but could not identify the exact flower species, reporting that the available magnification was not sufficient to reveal the needed microscopic structures.

The HuskyLens 2 handles image capture and includes its own onboard machine-vision features, including object tracking that, in one demo, locked onto and tracked a small insect that wandered into frame. The Jetson Orin Nano supplies the compute for the heavier Gemma model, letting the system run without a cloud connection. The creator also notes the current mount's lack of a refined focus-adjustment mechanism makes precise focus difficult, an optical and mechanical limitation rather than a compute one.

Beyond biological samples, the build proved useful for a more mundane task: inspecting tiny electronic components and reading markings on a Raspberry Pi Zero 2 W, a reminder that edge-AI microscopy projects can find practical value in workshop and inspection use cases even before their optics are lab-grade.

Hackster.io's report does not include benchmarked model performance or a stated roadmap from the creator; future iterations would need a better lens and a proper focus mechanism to move past general-observation-level identification toward more precise specimen analysis.