Raspberry Pi AI EYE Camera Uses AI

This Raspberry Pi project is really picture-perfect for AI enthusiasts and tech buffs. Maker and developer Oscar Wilmerding created this camera from scratch using SBC, the Raspberry Pi, with some added flair. He's thrown AI into the mix to take the images it captures and process them into something completely new. He's dubbed his new creations AI EYE.

The way the project works is simple, unlike the nitty gritty details of its construction. To use the camera, you take a picture of something. It then uses AI to scan the image and create a detailed picture description.

This description is then parsed to another system, creating a new picture from scratch based on the textual information. You can also use a descriptive word to give the new image a theme. For example, you can take a picture of a cat and have it recreate the photo with a spooky Halloween vibe.

Wilmerding was kind enough to make the project open source, so there's tons of information detailing exactly what hardware he's using and how you can recreate it at home. According to the official project page at GitHub, he also plans to make and sell kits for anyone who doesn't want to gather everything from scratch, but only if there are enough interested parties.

Wilmerding confirmed that the camera is built around a Raspberry Pi Zero 2 W. This makes for a good, portable option. Still, it lacks power, so the AI functions are handled using separate online services rather than locally on the Pi. The Pi Zero 2 W receives power from a 3.7V lithium-ion battery housed alongside the Pi in a custom 3D-printed shell. This shell also contains a custom PCB and a 3.5-inch LCD for real-time visual output.

Two AI systems power the project. Astica Vision creates text-based image descriptions. The second service is DALLE 3, which uses the image description data to generate new pictures.