Facial Recognition 2

So, the hard bit! How to make Scratch ask for a person’s facial details, then compare to its database and return a match?

Well done Riccardo & Kip for solving this. Here’s their project, click on the picture to go to the Scratch website and See Inside to see the scripts & try it out. Neat eh?

scratch R

 

 

 

 

 

 

I was particularly proud of Priya & Izzy this week because they really persevered when their project didn’t work for the second result. and then, just when they had finalised it, they decided to change the sprite and lost all their scripts (thanks lovely Twitter friends for trying to help them). I feel their pain as I did exactly the same recently! Anyway, it took them less than 10 minutes to rebuild their project which just goes to prove my point – “Learn slower, learn deeper” 🙂

Everyone else is SO nearly there, and people have tried different approaches. Some scripts are long but will still work and some have changed their code to tidy it up or make it more efficient. We learned that:

  • in order to simplify our coding, we could abstract unnecessary detail – we don’t have facial scanning & measuring software so we simplified the process by manually measuring & entering these details
  • although understanding & coding facial recognition seemed like an enormous task, we could decompose it into sections and begin to tackle it those sections more confidently – build a database, interrogate a database and return results.
  • we could debug our programs and improve them as we went along – for instance, at the beginning we thought we could jut input one measurement, then only put in the second if there were duplicate results. But we soon found out it was more efficient to ask for all three measurements at the start.
  • we can generalise our coding and understanding to help us develop other apps – now we know how to make a database we can apply this solution to other problems and don’t need to think all over again (see voice recognition below).

Today we also discussed how we might make Voice Recognition software. We discussed using the pattern of sound waves, teaching a computer phonic sounds, and lots of people recording the same word in different accents so that the computer can recognise the words. We also practised using voice recognition on Jon’s Android phone (Google voice recognition) to find the way from school (or current location) to Waterloo station. It worked perfectly! Our discussions then moved on to how GPS systems/ Sat Navs in cars worked and how they add in real-time traffic information – video camera data, monitoring speeds and even tracking the movement of passengers through their phone’s GPS signal.

We talked about the new Amazon device – “The Echo”, which is:

echo

“A 20cm-tall black cylinder, that sits in your home and listens to everything you do. You will want it to do this, if Amazon’s marketing is to be believed, because it will be able to answer questions like, “who is Abraham Lincoln?” and perform simple tasks such as adding ice cream to your shopping or playing a Taylor Swift track. (You have to say the “wake word”, “Alexa”, before it will act on what it hears)… The device is like a hyped-up Siri or Google Now for your whole house..It’s an always-on, Wifi-enabled obelisk listening to everything you say”

(From The Guardian G2 10th Nov 2014.)

Then (we were very chatty today!) we had an interesting chat about Smart TV’s and how they send you adverts and may collect data from you without consent – LG TV’s were found to be sharing information about what their users were watching, and also the names of files on any USB stick that owners plugged in – see this article:

lg

 

 

 

 

 

 

 

Jon showed us an app he had found last week which is similar to our idea – Leafsnap. You can take a photo of a leaf and it will try to tell you which tree it’s from. We discussed that this app cleverly “crowd sources” – it asks you for feedback on which leaf you think it is and then it uses your inputs to improve its own database. We thought that this would be a good idea for our app too. The more reliable data you collect, the more accurate your app will become. Big data!

leafsnap

 

 

 

 

 

 

Finally, Ollie had kindly brought in his magazine “How it Works”, to show us an interesting article about facial recognition & fingerprinting. I will put it up on display for everyone to read.

Well done everyone! All this in under 2 hrs!

4 thoughts on “Facial Recognition 2

  1. I know a good crowd sourcing app, it is called Maprika, I use it to find out where I am when I am skiing, it uses GPS, like Satnav, to show me where I have been and how far and fast I have been.
    But it didn’t originally come with any maps, people had to add their own maps with GPS co-ordinates and then they were available to everyone.
    You could draw your own map of the school and surrounding streets, upload it, “co-ordinate” it (ha ha – joke) and follow yourself around the school.

      

  2. When we were doing the project, I realized that when you are coding, it is basically the same as logic. You just have to say what exactly you are doing out loud. It helps a lot. That is how Issy and I managed to complete the project…almost. 🙁 I would recommend the rest of Code Club to do this if they are stuck.

      

Leave a Reply

Your email address will not be published. Required fields are marked *