ID your friends with a palm-sized DIY AI camera

Welcome to the brave new world of Artificial Intelligence (AI) photography.

What’s all of the hubbub about AI photography? Is it just another tool resting amongst the other clutter inside your camera bag of tricks? Would you like to try a teensy bit of AI photography to see if it has any merit in your workflow? Well, Espressif has your AI covered.

Specifically, Espressif has a ridiculously small AI development board that is called the ESP32-S3-EYE V.2.2 that’s priced at $45 and can be purchased from Mouser Electronics. Cracking the code on this brain-twisting naming convention we see that: ESP32 is the name of main processor on the board; S3 is a variant of the processor; EYE refers to the onboard 2-megapixel camera; and, V2.2 tells us that this product is the second major iteration of the Espressif AI development board. For simplicity’s sake, let’s just call this product “Evil EYE.”

Getting to know the Evil Eye

The amount of technology that has been packed into this 1.25-inch x 2.5-inch sliver of silicon is truly mind boggling. Naturally, starting with the 2MP camera, which serves up just enough resolution to make the Evil EYE into a powerful AI system, it’s very easy to overlook the numerous buttons, LEDs, and battery support circuitry that all contribute to this board being a discrete, portable photographer’s companion.

Add in the digital microphone, USB port, WiFi support, microSD card slot, and an attached sub board (Espressif calls this the ESP32-S3-EYE-SUB) that features a 1.3-inch LCD and you’d swear that the Evil EYE has everything you’d need for successfully experimenting with AI photography. Well, yes and no.

Sure you can explore the capabilities and possibilities of AI photography…that is, as long as you just want to study face detection and facial recognition using only your face! Huh? You see, Espressif, for some unknown design decision, implemented the LCD-bearing sub board on the same side of the Evil EYE development board that holds the camera (Please note: this decision might have been based on security/privacy requirements). Yes, the LCD sits about 0.25-inch below the 2MP camera. Therefore, during operation of the Evil EYE, you will only be able to see your face on the display. So much for innovation, eh?

There’s a better way to AI

You power the Evil EYE with a 5V supply that is connected to a micro-USB port. There is also an optional set of copper pads for attaching an external battery pack. Additionally, an on-board battery circuit is available for recharging a connected battery pack via the USB port.

Once power is connected to the USB port, you can use the Evil EYE for face detection and facial recognition. Optionally, using a portable power bank will convert the Evil EYE into a powerful and portable AI photography system.

After a short power-up, the Evil EYE LCD will display a remarkably clear video stream. And if the camera lens is pointed at you, your face should be surrounded by a floating blue box. This box indicates that a human face has been detected.

Pressing the MENU button on the Evil EYE circuit board initiates the facial recognition software. If your face is human, an ID will be displayed (e.g., 1, 2, 3, etc.) and your face is now captured by the camera/software. Using the UP+ button enables the Evil EYE to begin scanning for other people. Every time your captured face is displayed, the previously assigned ID will appear on the LCD. When an “unknown” person is detected by the camera, “WHO?” is displayed on the LCD. Your newfound “visual” friend can be assigned the next available ID with the MENU button.

Recognized face IDs can be deleted with the PLAY button. But that’s a crazy number of button presses for a circuit board that uses so many itty-bitty buttons. There’s gotta be a better way to controlling the Evil EYE. Indeed, that better way uses voice commands to operate every AI feature. Embarrassingly enough, you must utter, “Hi ESP” to enable voice command mode. Geesch!

Thank goodness, you only have to speak that wake-up phrase once. This mode remains active until you say, “Stop Working.” In order to get this conversation going, try saying, “Enter Face.” This command will give your face an ID…just like pressing that MENU button.

A thoughtful visual cue is provided for each voice command that is understood by Evil EYE by using a blink of its green LED. Furthermore, you have a whole vocabulary that you can verbally engage with your new one-eyed AI friend: “Recognize Face” and “Delete Face,” which displays an ID or WHO? and removes an ID, respectively. HA; not a single button press was required.

Voice commands can also be used for specifying operation modes: “Face Recognition,” “Motion Detection,” “Display Only, ” and “Stop Working.” While the Face Recognition mode is the normal AI operation for the Evil EYE, the optional Motion Detection mode reacts to movement “seen” by the camera by displaying a small blue rectangle in the upper left corner of the LCD. Alternatively, a stock video stream is shown on the LCD in the Display Only mode.

All-in-all, the Espressif ESP32-S3-EYE V.2.2 is a fun, albeit limited, introduction to AI photography that puts facial recognition into the “eye” of the beholder.

Enjoy.