I had the pleasure of speaking at the last LondonR event of 2020. What a strange year it has been? But this put the icing on the cake.
The premise of my talk was to take a novel Kaggle parasite cell dataset and advocate how this type of classification task could be transported to other areas such as clinical x-ray scanning, diagnostic image condition detection, etc.
The live event
To view the talk, have a look at the LondonR event below. I was on first and then two very interesting talks followed from Gwynn Sturdevant – FasteR coding: vectorizing computations in R and Stuart Lodge – Raindrops on roses and whiskers on kittens – a few small things that make me a happy R developer:
Where to get the content
The presentations from the session here. The GitHub code for the convolutional neural network can be found by clicking the GitHub button:
I have written a tutorial about this in my previous blog: https://hutsons-hacks.info/nhs-r-community-lightening-talk-computer-vision-classification-how-it-can-aid-clinicians-malaria-cell-case-study-with-r.
This presentation had two new addition, on top of what was presented recently at an NHS-R Community Conference event.
Face Mask Detector
The two new additions, delved into how computer vision classification can be used with localisation (bounding box) detection to create novel ideas such as a Face Mask Detector:
The Mango team were used, as an example, of how facial regonition – specifically the YOLO framework, can be used to detect faces in Python:
The code, for this Python file, is also contained in the LondonR file.
Enough of my ramble
This was a really fun experience and I would urge anyone with an R based project to sign up for LondonR. The hosts Mango are great and you will be treated to a feast of discussions, as well as really friendly people to boot.
Look out for my next blog post on this site and please follow the social icons below to connect with me.