Win a Raspberry Pi 4 in Celebration of Our 10th Birthday!

Learnt AI and ML but Never implemented any AI/ML model on hardware.

Its time to implement it on real hardware.

Yesterday saw an article on tensor flow lite implementation on ATSAMS51 from ladyada. It was amazing.

I will use RPi4 to learn and implement ML/AI and build some easy to do examples. Because I know there are lots of other people like me who have learnt ML/AI but never did any projects or want to learn how to build a project using ML/AI.

Riding bikes who will not love them? But a recent worst riding experience of mine gave a new project idea.

Problem:

We all love travelling to new places. While using bikes, when we travel to new places, we tend to use GPS based maps and mostly it will be our smartphones with Google Maps. Those smartphones will be placed in the handle bar. So that, it will be easy to watch and drive. Some prefer to use the audio output from GMaps but I don’t recommend it because hearing near by vehicles horn/ any important sound will be affected.

Once I was travelling to a new place in bike at night time, it started raining heavily. Since I wear spectacles it made the situation worse. I had a poor vision of the road and traffic. But I still continued since I had to reach back home. Since it was a new route I had to stop, take my phone out and check the routes. Since I don’t want to drench my phone despite it being water proof, I did not place it in the handle bar. Stopping and checking the routes was also tough since the path I was travelling was a highway with no place to shelter. So with wet fingers and display I had to check the route. Also, l missed some important phone calls and SMS since my phone was in my pocket after I had checked the routes. Another problem was since I had a poor vision of the traffic in front of me, I might have hit them. Luckily without any damage I managed to reach back home safely.

Solution:

The Project which came to my mind after this worst riding experience of mine is explained below. I don’t know whether this is already available in the market for bikes but I bet they would be costly if they do exist.

The solution would be to make a water proof display system that can be mounted in the handle bar. I already have a RPi 3B and 7 inch touch screen display. I have experienced lags while using the combo as a tablet. But lag in critical real time situations can make the situations worse. So Raspberry Pi 4 would be a better alternative. The system will therefore have RPi 4 and 7 inch touch screen display with battery pack to power it and placed inside a water proof case. This system will be able to

1. Display Google Maps and we can navigate using it.
2. Connected to phone so that we can get an alert when a call / sms arrives.
3. Sense traffic and alert when we are very near to the vehicle in front and about to collide.

Dixon Selvan

I would like to use the new RPi 4 with a software designed radio (SDR) I picked up a couple of years ago.

With its extra horsepower I think it could produce some interesting analysis of radio spectrum.

DAB

I'd love to use the new Raspberry 4 to build a security system around facial recognition and a 2-factor authentication. The proof of concept would involve unlocking a door with "familiar" faces and a cellular-based confirmation mechanism to avoid false alarms and to reduce security flaws -like trying to unlock it with a picture-.

A project like this may take advantage of the improved processing power of the new RPi 4B and would be a good opportunity to test technologies like the Hologram Nova with the new member of the Raspberry Pi family.

Luis

I just came home from a month on a research vessel in the Indian Ocean.  One of our projects is a stabilization platform to use a scanning doppler Lidar on the ship.  The system is similar to a camera gimbal on a larger scale.  The system we designed uses a raspberry Pi 3 as the platform controller.  The next version will most likely use a Pi 4.  Here's a pic of the control box:

I'll upload a video of the system in operation on the ship.

Scott

I would like to make a  Attendance  tracking system along with check-in and check-out time using camera vision on the Raspberry Pi 4 along with the camera (Kuman Noir 1080 p with focus ) which I have , so that this can be project which can be implemented in schools and colleges  for automatic face recognition based Attendance , then I also aim to work on Machine Learning based projects using the Raspberry Pi 4 which has 4 Gb and would be helpful for making AI based projects and can have wide range of implementation.

Thanks

Aabhas Senapati

I have a number of projects that have been on the shelf waiting for the right SBC, time and motivation. The Raspberry Pi 4 sound like a good fit for some of them.  Here's a couple.

1. A WiFi access point - yes I know there's lots of these and easy to do with any RPi but what I want is the performance of GBit Ethernet and memory + CPU performance to also run intrusion detection software and also act as a VPN gateway server.

2. General purpose portable desktop - A Laptop by any other name - I started this with the recovered parts from an old HP laptop. I have built this so far and had it working with RPi3 but it's performance isn't quite there ... but I am now thinking of how I could incorporate dual displays as the RPi4 supports that, maybe a fold out side panel?

I would rather like to dive into artificial intelligence.

Application would be something like weed detection.

(c) Olle Welin https://github.com/ollewelin/Object-detection-Dataset-Crop-Weed2

Remember my [Pi IoT] Plant Health Camera in which I displayed a vegetation index, using opencv.

I also did something similar by extending it to augmented reality glasses (https://precisietuinbouw.nl/Portals/39/Projectdossiers/Poster_Patterns_For_Profit.pdf?ver=2018-08-08-142121-593 ) in which the Pi as WiFi and webserver streamed the NDVI images to android based AR glasses.

I hope the Pi 4 is powerful enough to run a convolutional neural network such as tensor-flow or pre-trained networks like R-CNN.

I would like to try, how well would the new Pi handle installation and running of a Gitlab server. Although most of professional developers use version control systems at work, many hobbyists and makers usually go with the "zip to archive" path.

I use Gitlab only as an end user at work and it would be great to try running a home server for managing code versioning, documentation (wiki style), maybe even backlogs with tasks for personal projects. The advantage of a distributed version control system is also that more people can work on things independently.

I don't think I currently have a project to be checked / build / deployed automatically, but Gitlab is also very strong in supporting continuous integration and continuous delivery.

I think the extra power would work well for things like image recognition, using OpenCV.

Having been inspired by Clem and Andy's Dragon detector, and encouraged by John Clift, I'd like to see if I can make it detect windmills.

clem57, Workshopshed, jc2048

Cheers,

-Nico