Keras Image and Video Convolutional Layer Neural Network.

So I tried real simple text based image analysis, but the reality is that images are 24 bit rgb style and Tensorflowjs with it's floating point math can handle the larger amount of data.

So lets try to do something similar to the last page and train the model with full images loaded from this web domain REMEMBER: THIS IS PROOF OF CONCEPT, IT WILL NEED MANY MORE IMAGES TO GET GOOD RESULTS!

Test page for loading multiple images of different sizes, force them to 64 x 48, and then load the data into a tensor as the xTrainingData for a Neural Network.
(Labels would have to be dealt with using a coordinated text file!)

https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/soccer-ball-0.jpg, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/soccer-ball-9.jpg, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/random/ran3.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/random/ran4.jpg, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle01.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle02.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle03.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle04.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle05.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/orange.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/white.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/rectangle01.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/rectangle02.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/rectangle03.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/rectangle04.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/circle-rectangle.png, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/random/ran1.jpg, https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/random/ran2.jpg
Labels are one hot encoded no shape, circle, triangle, rectangle
0,1,0,0, 0,1,0,0, 1,0,0,0, 1,0,0,0, 0,1,0,0, 0,1,0,0, 0,1,0,0, 0,1,0,0, 0,1,0,0, 0,1,0,0, 1,0,0,0, 1,0,0,0, 0,0,0,1, 0,0,0,1, 0,0,0,1, 0,0,0,1, 0,1,0,0, 0,1,0,1

All images below fixed 100 x 100 and pixelated if low resolution



Test or change this data
https://hpssjellis.github.io/beginner-tensorflowjs-examples-in-javascript/saved-models/soccer/ball02.png

Learning Rate:
BatchSize:
Epochs:
Reshaped image -->

Rear Mobile Camera Front Mobile Camera
Take snapshot every milliseconds

...