ml5.js: Classifying Drawings with DoodleNet

[@Roknok]

You cannot explain how happy I am when you upload a video (which is awesome)

[@rashidsiddiqui4502]

Great project idea

[@yoogikovendhan6802]

This is really good!! Also your video series on neural networks inspired me to create my own from scratch in p5.js

[@neillunavat]

Im 14 years old and it inspired me too! I did it in python and I also did a for loop for n number of hidden layers! Ahh im so grateful because people like Dan exist :))

[@purshotambohra2131]

@@neillunavat Wow at age 14 you are doing such things, congrats! Never knew people are getting that intelligent to pick up programming so quickly. Now you would have been 16 years old, and I am curious about your progress so far in these two years.

[@thehappycoder3760]

watching this while on the cross trainer at home in lockdown UK. Getting fit and educumacated at the same time

[@vidhanp482]

you are a godlike teacher

[@AleksanderFimreite]

Possible answer to question asked at 3:27... Could the warning somehow be related to the color value of the pixels? Receives 16, expects 1 kind of sounds like it is looking for only fully black or fully white pixels. And the reason its only a warning is probably because they are doing some post-processing simular to that of the treshold filter behind the scenes.

[@nkpvg6805]

Really Epic Video As Always Keep It Up Dan

[@rajatguptasupercoder9789]

Best smiling teacher I wish you would have taught Python

[@RajJaiswal538]

Javascript is good too :)

[@sajeevanfuels1980]

Daniel Sir Can you tell me which book should I prefer for javascript coding that includes all the concepts right from variables to physics enjine and everything please tell! :)

[@petaca39]

hello good explanation a query can generate a PWA (progressive web application), using ml5 and p5.js. If it is possible to do it, you would have a link with information on how to achieve it, it would be enough with this same example of the video but in PWA format.

[@Ping2001]

Can this model be inverted? Giving a category and let it draw?

[@hhhharis622]

That would be so cool to see

[@papesldjnsjkfjsn]

no, classifiers can't do that, but can train your own model

[@TheCodingTrain]

Take a look at my video on sketchRNN! kzbin.info/www/bejne/kHS7nKmtgd11qKM

[@erikjohnson9112]

3:26 GUESS: Shape input tensor mismatch. The only difference is 16 vs 1. My guess, the image is 16 bits/pixel and model expects 1 bit per pixel.

[@Strak3r007]

That has to be the reason, the trained data confirms it.

[@EeVennSoh]

This is awesome!

[@اوراق-ح7ه]

what's about graphice tablet ..can i use the pen pressure

[@வணக்கம்-ங1ண]

Thanks Dan. 💛💙💜💛💙💜💚

[@Beast80K]

Sir I have a doubt , I created a classifier by watching your videos I want import all data set for train using input file tag of html . Would you please make a video on this ? -Thank You your videos are very helpful

[@Anujkumar-my1wi]

can you tell me what those hidden layer are doing and also how increasing the number of layer can increase the complexity of the feature and why do we have to use nonlinear transformation can't we just use n number of hidden nodes with linear activation function?

[@sergiorey2049]

Awesome videos

[@faiththacreator6678]

Hi, I wanted to know how to do Coding for a Game? For example, making the player cook!

[@bijoyroy6112]

Please make a video on Decision tree and pruning

[@jancassio]

Daniel, the reason why you're getting this warning is because TF2 expects an 28x28 uint array which each index should be 1 for filled pixel or 0 for empty pixel.

[@Anatoli_Derese]

What is the text editor he is using?

[@thelavagod]

p5js.org

[@TheCodingTrain]

It's editor.p5js.org/ !

[@jjte3921]

Can you make video about Lua. Language?

[@gaurishagrawal9404]

Is this pre-trained DoodleNET available for python?

[@guyincognito1985]

Hopefully there will be more ml5 videos to come...?! 🐱🍓⛄

[@faiththacreator6678]

Also does Coding have math in it I’m not really good at it

[@RajJaiswal538]

Yup

[@eonurk]

Can we just upload this code to github and run it on there by setting the docs? Simple game/website to share without paying anything at all? Btw are you attending to any seminars or something in NYC in near future, I would love to meet Dan! Huge fan, love your videos 😄

[@lalropekkhawbung]

Cool bro

[@ghulam2545]

Cool

[@geoffwagner4935]

What's really astonishing is this is all getting any results at all with a webcam they are so cheap and noisy, it really just be the nature of the whole thing "getting everything as close" as it can that it gets anything at all from webcam pixels. For Metashape i wouldn't even dare. i know these pixels are so noisy and so not like each other at all pixel by pixel. image alignment you only get a few speckles that match up out of 100s of photos, you get a good Sony and it practically will make a perfect model just from image alignment and can really do it with very few photos, if they're good. i can't wait to get a better adapter for a good camera .

[@aditioswal]

Wow!!!!

[@dabll0236]

is this a kids show?

[@rujor]

That number formatter does a really, really, poor job 😄...

[@thelavagod]

early

[@sarvagyasingh4518]

tell me a thing this guy cannot do in JavaScript