Raspberry Pi AI Camera Quick-Start Guide

[@shakejones]

brilliant! best setup video on youtube, well done mate 🙏🏻

[@iCottage]

I really appreciate the quick and snappy tutorials that get right to the point ❤

[@falxonPSN]

Short, sweet, to the point, and DAMN useful. This is how all tutorials should be!

[@tarunjeetsingh1111]

Loved the video!!

[@hehehehagrrrr1319]

Can you make a tut on how to train yolo with my own custom dataset in rpi 5?

[@Core-Electronics]

Training your own models is quite an intensive process both time-wise and processing wise. It can take an RTX 4080 up to 5-6 hours to train a decent sized model, and without a high end GPU it can take days. We are unsure if we have a guide coming in the future, but its in our list of ideas. In the mean time though, Ultralytics has some fantastic documentation on how to train your own model: docs.ultralytics.com/modes/train/#key-features-of-train-mode And they even have their own HUB service where you can rent a GPU to train the model for cheap: docs.ultralytics.com/hub/

[@RandoomFlash]

Is this camera works for facical landmarks recognition?

[@jackdeidolci]

Very nice! Do you think that this AI camera alone has got enough processing power or would/could it benefit from the additional Ai HAT+ (Hailo5) module? I'd like to use it for posture estimation but it has to be good.

[@Core-Electronics]

IT definitely has the capability, using the AI HAT+ is currently a little difficult software wise. We have also just made a pose estimation video running without the AI camera or AI HAT and we managed to get some decent FPS! kzbin.info/www/bejne/gam8lHScqt53pcU

[@tjones99]

Wondering if you think AI camera and OpenCV would be capable of detecting a water skier. There is a lot going on in the scene with spray, changing position, wakes, and background shore images and the skier moves at pretty high speed. Would it be possible to train it to recognize a skier with all that going on?

[@Core-Electronics]

The AI camera is designed to be the less powerful but easier-to-use AI accelerator. If you need something a bit better, you might want to check out the AI HAT: kzbin.info/www/bejne/kJnXY3pnndl5fqM In terms of recognising a water skier, you would probably be able to detect people in general with it, but you wouldn't be able to detect the difference between a regular person, and a person who is skiing. To train a custom model is quite in involved process and from a rough guess it may be difficult to train it to recognise the difference between a skier and a non-skier if that's what you are after. The other option is to use a pre-trained model that someone else has made, but I think you may be one of the first people in the world to use computer vision for skiing! If you are looking to throw together a project and need a hand with it, we have a maker community forum if you are interested: forum.core-electronics.com.au Best of luck!

[@tarunjeetsingh1111]

Please make a video on pose detection with raspberry pi 5 and camera module 3!!

[@Core-Electronics]

Coming soon!

[@weihong8337]

waiting for you

[@BryanMcAllister-f4t]

does the camera come with the cable?

[@Core-Electronics]

Sorry for the late reply, yes it does!

[@Errorcutive]

Hot take on This vs Jetson ?. pro and cons

[@Core-Electronics]

Probably a more apples to pears comparison here, this is moreso designed to add some "easy to setup and get going" machine vision to any raspberry pi, and its just the right amount of power for basic computer vision needs. The Jetsons on the other hand are more professional-tier platforms to develop and deploy your computer vision or AI application. This would be no where near as powerful as a Jetson, but its much cheaper and uses significantly less power.

[@Pete-d3j]

im not computer smart but I want something like this connected to a translator app on my phone.

[@Mr.RobotKid]

Imagine using this in a robot... So good tech.

[@Core-Electronics]

This is just the start of running computer vision locally as well! Exciting to think what will be possible in 5-10 years!

[@imdefnotash]

70$ 😵

[@Mr.woman_lover]

take my advice. that hair aint working bro

[@Core-Electronics]

We've taken your remarks under advisement and we will be shaving Jaryd bald before the next video comes out. Thank you for your assistance in this matter.

[@iceprada2]

😂😂😂😂😂

[@iceprada2]

@@Core-Electronics😂😂😂😂😂😂😂😂

[@StickyWeiner]

@@Core-Electronicsplease do a reverse Mohawk🙏