Charmed Labs LLC Pixy2 Smart Vision Sensor - Object Tracking Camera for Arduino, Raspberry Pi, BeagleBone Black

The Pixy2 is a easy to use and ready to run smart camera for feature detection and learning capabilities.It is a smart camera that processes image data on board at high speed.Bandwidth for the communication to the supporting processors (Arduino, RPi, BeagleBone, etc.)is minimal due to the transmission of abstracted information rather than the complete feed.It is possible through special "tricks" to access also the raw image feed, however, the concept is to processdata on board of the camera rather than on the processor MCU. I assume future directions of the Pixy will gointo further increasing flexibility of programming of the camera and its intelligence, which could leadto an entirely new style of programming altogether. I would be looking forward to see the Pixy X in the future.

This is a great entry into the 'add vision to my project' world. Out of the box this guy detects coloured objects, assigns them a name allows you to feed data to your microcontroller project - where you can do whatever you like with that project.. Like have you robot follow said-object around the place. As a company they seem really friendly and i'm hopeful they are responsive to questions - as i have a few. Overall though, it seems like a great bit of kit at a reasonable price.

Absolutely love it.I used this with my quadrotor for object tracking and potentially line following.I connected it to my Beaglebone Blue through UART using the libraries that they provided for Arduino. I basically just modified the Serial.read() function to the one that I use for reading.Overall great product, great stability on the identified objects. Pretty powerful little cameras.

Well, it's a brill idea, but I just couldn't get it to work 100% on my Raspberry Pi 3B running Python 3.5, and it's also a bit expensive, so 3 stars for idea only.

Very good product which easily connects to the Arduino.Also very light weight and compact, so is ideal for our robot. Would recommend for track following robots, where there may be a need to anticipate a change in the track before arriving at that change.