How Well Do Machines See?
Tensorflow.js Machine Learning Experiments
I decided to get under the hood of machine learning (ML) and created these web-based prototypes. These prototypes offer low latency responses because the ML models used are optimised for web environments. Give them a go and have fun! (NB: Prototypes 1, 2, 4 and 5 work on desktops and mobiles, but No. 3 will only work on mobiles.)
Give them a go!
Tensorflow.js - static image classifier
https://tensorflow-image-array-classifier.s3.eu-west-2.amazonaws.com/index.html
Tensorflow.js - webcam image classifier
https://tensorflow-webcam-image-classifier.s3.eu-west-2.amazonaws.com/index1.htmlTensorflow.js - smartphone rear camera image classifier (see images above)
https://tensorflow-rear-webcam-image-classifier.s3.eu-west-2.amazonaws.com/index.htmlTensorflow.js - this webapp lets you train your own model
https://tensorflow-transfer-learning-knn.s3.eu-west-2.amazonaws.com/index.html
Tensorflow.js - I trained a model to recognise images with Teachable Machine and deployed it
https://tensorflow-trained-teachable-machine-getusermedia.s3.eu-west-2.amazonaws.com/index.html