...
...
Web Page | Explanation | Video | CodePen, Glitch | Deprecated at Version |
Tensorflowjs Version 1.0.0 and newer |
tfjs01-functions.html | Updated to also show functions with multiple variable returns using arrays and objects and module solutions using namespacing | Video | .... | Active Version |
tfjs02-basics.html | Beware lots of Tensors in one function | Video | .... | Active Version |
tfjs03-general.html | General Tensors and how to use them | Video | .... | Active Version |
tfjs04-quadratic.html | First Neural Network solving the quadratic equation | Video | .... | Active Version |
tfjs05-matrix-math.html | Tensors and Matrix Math | Video | .... | Active Version |
tfjs06-javascript.html | A reminder of many things Javascript can do. This is still in draft mode. I have a lot still to do here | ... | .... | Active Version |
tfjs07-xor-neural-network.html | Solving an XOR Neural Network | Video | .... | Active Version |
tfjs08-knn.html | K-Nearest Neighbor Machine Learning Classification | Video | .... | Active Version |
tfjs10-xor-layers.html | Bit slow but very easy to read the layers. Original by Matt Cameron. His Code is here. | ... | .... | Active Version |
tfjs11-load-weights.html | Finally got weight saving and loading working | ... | .... | Active Version |
tfjs14-sinusoidal.html | I just really like sinusoidals | ... | .... | Active Version |
20keras-xOr.html | Finally made a system for working with Keras Layers. This is the easiest example | ... | .... | Active Version |
21keras-integers.html | Now Keras Layers using Integer inputs and outputs | ... | .... | Active Version |
22keras-decimal.html | Now Keras Layers using float (Decimal) inputs and outputs | ... | .... | Active Version |
23keras-digital.html | Example IoT device 8 bit input 4 bit output Smart Garden example | ... | .... | Active Version |
24keras-one-hot.html | Why use one-hot encoding | ... | .... | Active Version |
26keras-small-image.html | Loading really small black and white images using a convolutional Keras Layer | ... | .... | Active Version |
27keras-medium-image.html | Very cool, takes any image and converts it to a 32 x 32 single digit text image, 0=white, 1=black, 2-8 ROY-G-BIV 9=brown. | ... | .... | Active Version |
28keras-full-image.html | Now much more difficult. Full RGB color 64 x 48 pixel images, loaded from the web (CORS OK) with webcam analysis | ... | .... | Active Version |
pre-trained/31-squeezenet.html | Finally got this very cool Squeezenet webcam demo working.
I didn't have the image data normalized between 0 and 255 | ... | .... | Active Version |
pre-trained/32-mobilenet.html | Finally got this very cool Mobilenet webcam demo working.
I didn't have the image data normalized between -1 and 1 | ... | .... | Active Version |
pre-trained/33-mnist.html | MNIST white numbers on a black background | ... | .... | Active Version |
pre-trained/34-switch-models.html | Model switching example. Must first find a "4" then trapped on the second model until you find an "8". Note: must be white numbers on a black background | ... | .... | Active Version |
advanced-keras/40keras-lstm.html | An LSTM attempt presently not optimzed for large inputs | ... | .... | Active Version |
advanced-keras/41b-cloud-gpu-challenge.html | My Cloud GPU TPU challenge. this is the new version. It has a link to the old version | ... | .... | Active Version |
advanced-keras/42-knn-classifier.html | Very cool KNN-Classifier example that you can easily train in the browser. Just waiting for how to save and load the data | ... | .... | Active Version |
advanced-keras/43keras-multi-input.html | For a while I have wanted to start making Advanced Keras Models. This is the first. It has two inputs and one output, unlike a regular sequential model that always has one input and one output | ... | .... | Active Version |
advanced-keras/face/44-face-knn-classifier.html | This is a KNN-Classifier connected to a face-api.js Should be able to identify your friends. Waiting for saving the knn-Classifier | ... | .... | Active Version |
advanced-keras/face/45-face-classifier-saving.html | Made this a while back and forgot to put it up. Shows how to save and load a knn classifier, by tricking tfjs to think the classifier is a multilayer tfjs model. | ... | .... | Active Version |
advanced-keras/46keras-multi-output.html | This has one input and two outputs. | ... | .... | Active Version |
47 Webcam face-api.js for Mobile or desktop for beginners | Github here | A must lookat Face Landmark detection webpage | On CodePen.io, On Glitch.com | ..... |
interpretability/50-interpret-grey.html | Sheding light on the black box | Using black "1" and white "2" to make shades of grey then using interpretabilty to analyse the predictions using "0" for transparent, which the neural network has not been trained on. | .... | Active Version |
beginner-examples/tfjs51-activation-functions.html | Testing making your own activation functions. not yet working | ... | .... | ..... |
. | .. | ... | .... | ..... |
. | .. | ... | .... | ..... |
. | .. | ... | .... | ..... |
Web Page | Explanation | Video | CodePen, Glitch | Deprecated at Version |
Some General Stuff I am working on |
My CodePen.io Tensorflowjs collection | I have made a collection of examples in both CodePen and Glitch. | Trying tofigure out which online IDE I like Best. | Presently leaning towards CodePen for Front End and Glitch for Front and Backend in node. | ..... |
My Glitch.com Tensorflowjs collection | I have made a collection of examples in both CodePen and Glitch. | Trying tofigure out which online IDE I like Best. | Presently leaning towards CodePen for Front End and Glitch for Front and Backend in node. | ..... |
DenseNet121/model.json | A fully useable DenseNet121 Model with shard files in Keras Layers style made ready for Tensorflowjs | This means you can edit it, add layers, freeze layers etc, much more powerful than taking a model from Tensorflow which is a frozen model. | You may use these files. I have others just the download sizes start to get a bit big. | ..... |
MobileNet/model.json | A fully useable MobileNet Model with shard files in Keras Layers style made ready for Tensorflowjs | This means you can edit it, add layers, freeze layers etc, much more powerful than taking a model from Tensorflow which is a frozen model. | You may use these files. I have others just the download sizes start to get a bit big. | ..... |
MobileNetV2/model.json | A fully useable MobileNetV2 Model with shard files in Keras Layers style made ready for Tensorflowjs | This means you can edit it, add layers, freeze layers etc, much more powerful than taking a model from Tensorflow which is a frozen model. | You may use these files. I have others just the download sizes start to get a bit big. | ..... |
NASNetMobile/model.json | A fully useable NASNetMobile Model with shard files in Keras Layers style made ready for Tensorflowjs | This means you can edit it, add layers, freeze layers etc, much more powerful than taking a model from Tensorflow which is a frozen model. | You may use these files. I have others just the download sizes start to get a bit big. | ..... |
Machine2learn | Powerful online Keras model generator and full app development. My test program showing a generated model can be loaded | ... | .... | ..... |
Parallel Browser Training | Can't really show a demo, but you could install this nodejs server and have multiple browsers help train individual layers | ... | .... | ..... |
Loading 10000 of the 50,000 CIFAR-10 images | Same as the below one except this uses the .png files and actually works | ... | .... | ..... |
Loading 50,000 CIFAR-10 images | NOT REALLY WORKING Takes about a min to load the first 10 K images, presently colors are messed up | ... | .... | ..... |
Rocksetta version of TFJS Getting Started | My version of the Tensorflowjs Getting Started. Can edit and run even from a cell phone | ... | .... | Active Version |
Rocksetta version of TFJS Mobinet | Edit online even from a Mobile device | ... | .... | Active Version |
android-issue.html | Using the Addition RNN. I have included the ability to adjust the learning rate to show the issue with android phone. Try a learning rate of 0.01 for no NAN loss. The easiest version of the issue many Android phones are having using Tensorflowjs. Looks like it is part of the loss caculation with Layers. | ... | ... | Active Version |
kickstarter.html | Just having fun with Vega-Lite graphs and my New Kickstarter. | ... | .... | Active Version |
PoseNet Web Cam to BVH | Web cam working not yet converting to BVH animation format | ... | Code | Active Version |
PoseNet Security Camera | Proof of concept. Using PoseNet to start or stop web cam recording | ... | Code | Active Version |
TensorHtmlPrint.html | Prints any dimension Tensor onto a web page. Not very well written but does seem to work | ... | .... | Active Version |
Submarineering-knn.html | His original tweet is an interesting take on running Javascript. This page is the IRIS Dataset KNN to identify flowers based on Petal Length etc. | ... | .... | From a tweet by @Submarineering |
Face API for the Big Bang characters (Proof of concept) | .. | Original github at Code | .... | ..... |
keras-rpi3.html | Most simple webpage for testing if a device like a Raspberry Pi 3 can use Tensorflowjs | ... | .... | ..... |
tfjs14-sinusoidal.html | I just really like sinusoidals | ... | .... | Active Version |
tfjs16-console.html | For that one in a 1000 time when you need to test something in the console | Paste the code without the script tags into the browser console. | Some fancy stuff here that is interesting to checkout | tfjs16b-console.html Even easier example of using the console |
keras-links.html | So I am having trouble learning Keras. This page is my attempt to get many examples in one location so that I can study them and find the easiest way to teach it | ... | .... | ..... |
Codepen about video stream and image taking | .. | ... | .... | ..... |
My CodePen of loading lots of image files into a tensor | CodePen of loading groups of image files | ... | .... | ..... |
My Codepen about converting RGB to a 10 digit number | .. | ... | .... | ..... |
. | .. | ... | .... | ..... |
. | .. | ... | .... | ..... |