Teachable Machine 1: Image Classification

[@ceciljoel9577]

If the model is seeing your face everytime along with the object then wont it fail to give an accurate answer if you dont show your face with the object

[@geoffwagner4935]

truly some of the coolest stuff ever.

[@sandro-nigris]

Talented presenter. I subscribed to the channel.

[@gohelboy]

The way you teaching is very intresting but please make video on python base ml that would be amazing ♥️

[@ELHAUKEZ]

You need to train it for an idle stance. just train it to recognize just "Dan" so every other mode will just pop up when it should rather then it detects the most "Dan" in any of the other features.

[@வணக்கம்-ங1ண]

I love your videos very much Dan. 💛💙💜💚❤

[@geoffwagner4935]

incredible

[@Frankslaboratory]

Yay another video

[@Tim-Jaeger]

I think I know what I'm going to do this weekend :D

[@jamiekazemier3652]

Uit welke provincie kom je? Miss heb ik wel iets met robots voor je gerelateerd aan je video

[@Tim-Jaeger]

@@jamiekazemier3652 ik ben van antwerpen, jij?

[@luukvink4144]

@@Tim-Jaeger enschede hiero

[@jamiekazemier3652]

@@Tim-Jaeger friesland :p ik dacht dat je de first robotics competition miss wel leuk zal vinden, Google het maar eens

[@Tim-Jaeger]

@@jamiekazemier3652 het lijkt zo leuk, ik zou zo graag meedoen, alleen zijn er een paar problemen: Allereerst ben ik 18 jaar en word ik 19 jaar in april (dus het is voor mij niet eens duidelijk of ik mag deelnemen). Dan heb ik ook een team nodig, wat ik niet echt heb. En ten slotte lijkt het duur....

[@rohanthube2920]

Awesome!!!!!!!

[@BarracudaSd]

Thanks for sharing!!. You are a genius.

[@zingnolan]

Love you always

[@grainfrizz]

Wow. I'm a guy who works at a cyber security company. I wish I could give it tonnes of logs that are threat and not threat and I'll just send logs to it. :D

[@senshi33000]

Your videos are very playful. Thanks for this. I have a question ? How to add training without redoing everything?

[@YouDJeuR]

Very nice work ! 😀 About the detection of the train, I was wondering if it's detection was not due to the green screen that is taking most space of the camera ? Maybe I missed the part when you said it's transluscent ...

[@athukurianilkumar5352]

what about un-trained object show and what is output from model ?

[@beron_the_colossus]

Yay Machine Learning!

[@rikahirdawati8926]

please help me sir, what algorithm did teachable machine use to create that model?

[@AB-ud3in]

What if I want to run two models at the same time, one for image recognition and one for audio? Can I just do the exact same thing you did here twice for the two models? I’ve tried doing that but it seems they are conflicting and causing an error with WebGL that makes my whole sketch crash

[@khaledbensaad]

i have error when running the p5js without adding the (modelURL+ 'model.json') in the function preload , now it is fonctionnel correctly with my model.

[@TheCodingTrain]

If you'd like support come hop into the discord! thecodingtrain.com/discord

[@nordicview4837]

Thank you so much, Khaled! This comment helped me to run the code!

[@aliibrahim4873]

Hi Thank you for all your efforts, How can I determine the architecture of this network? Regards Ali A. Ibrahim

[@uzair004]

Can i we use this to identify different people using their faces?

[@masha.nazeem1615]

Hi, what if I want to add pose as input and get audio output?

[@geoffwagner4935]

wow, OO Nyu ml5 is literally in googles documents?

[@L76videos]

How to classify a 'nothing' image? If I set in example 2 images, the classifier gives me always 2 options, but using the web camera you watch also other objects not in the list of images classified. Thanks

[@ivanbezerra3531]

I love you

[@Its.abhi..]

10:26 is it necessary to import p5.js library also?

[@geoffwagner4935]

what is all this? these seam like such side projects, then Shiffman or ml5 is in something crazy official like this, in a google doc, or a processing one, i'm sure theirs more i'm not remembering it' shocking where these nyu names keep popping up. like "this is so unofficial i don't know what I'm doing" then there's google references just to your projects .

[@phvaguiar]

Export is not working! In p5 Uncaught SyntaxError: Unexpected token '

[@TheSaintsVEVO]

You must have a typo in the code. Probably missing a parenthesis is my guess

[@tzisorey]

I'm having visions of a mobile app that tells you what Hogwarts house the photo you've given it, most represents.

[@apollo8030]

when I do It it comes up with a type error: Failed to fetch The webcam still works but the console doesn't come up with the objects. Can anyone help me?

[@alonsosa8272]

Hi! The url to your model might not be the right one. Try this: Open a browser tab and go to the url that starts with teachablemachine.withgoogle.com/models/ Click on the “model.json” button. That will take you to a url that looks something like this: storage.googleapis.com/tm-model/higuys/model.json That’s the url you want to pass to ml5.imageClassifier(). It’s basically a json file hosted on Google Cloud Storage.

[@uhsperspective]

It gave me a different format for the link than we he has in the video. Mine looked like this: teachablemachine.withgoogle.com/models/xxxxxxxx/. If you have the same thing, try adding model.json to the end. That fixed it for me. (teachablemachine.withgoogle.com/models/xxxxxxxx/model.json)

[@tahasharjeel]

@@alonsosa8272 thank you so much!!!

[@ahsensaaim3692]

@@uhsperspective Thank you so much!

[@IsaacAsante17]

Please release the "Don't switch tabs" song! Awesome intro video by the way!

[@UserUnknown07]

That would be great 🤩

[@JawwadRafiq]

It would be really awesome. please sing it for us.

[@faroe2k9]

I wish i was 13 again and had you as a teacher in programming BASIC and ASSEMBLER!! I submit to your channel from this one video!

[@rahulkumarjha2404]

You are one of the most enthusiastic KZbinr out there who loves to teach.

[@arifulislamleeton]

Hi I'm Ariful Islam leeton im software engineer and Apps development

[@tinkumonikalita2576]

Great video Dan. Just watched the video by Google and then saw yours, and I clicked.

[@UserUnknown07]

Same here, yay! 😀

[@antonioriz]

Please help! I continue to receive the following error message: Uncaught TypeError: Cannot read property 'split' of undefined (: line 58); and some time the line 58 not even exist; who knows how to handle this? Thank you

[@lionpersia]

Getting this happy from very simple things? You're a very luck man, Dan, you have solved the mystery of the universe.

[@phptaskforce]

Hi, When I run the example from p5js web editor: editor.p5js.org/codingtrain/sketches/PoZXqbu4v I get this error *Failed to link vertex and fragment shaders.*

[@AlexGuerrero007]

Same problem, however only on Chrome latest

[@TheCodingTrain]

Check and see if you have WEBGL disabled in the browser?

[@phptaskforce]

@@TheCodingTrain Hi Daniel, thanks for a quick response. I had my Chrome on default settings and seems that WebGL was enabled but I still received the same error. But after I changed this setting *#use**-angle* in *chrome://flags* to OpenGL from default.. error disappeared and everything works now :)

[@B0bbySneeze]

Love the videos, you're the best ever!!! So awesome literally watched them all over the last month. #suggestion Could you please please please do a video on OCR, like if you input an image of a receipt, the output would be text. Wondering if Teachable Machine can handle this, going to try now! Would love to hear your thoughts

[@leonardsmith9870]

I've been programming for a few years now, and I'm convinced that this kind of stuff is pure magic.

[@aleexwh]

Is it weird that i am watching this, even though i don't know how to code?

[@TheCodingTrain]

Not at all!

[@transolve9726]

because he makes it fun, like a kids show. actually very powerful information in this video

[@sunshinelabz]

One of the best way I have seen some one can make complex tech so easy to learn, explore...love the enthu and art to demo...and showcase things that one can learn ... Keep going Sir... Love to see more and more new technology videos from you...

[@shubhrashukla7879]

Oh my god the new videos have SUCH good editing it's soooo good!!!

[@randalwhitmore9367]

This is one of the best coding tutorials I’ve ever seen. Looking forward to trying this out.

[@yifanli9950]

Actrually when I'm trying to use my own URL, it shows an error that "Uncaught TypeError: Cannot read property 'split' of undefined (: line 58)". I have no idea about how to solve it...

[@MultiMunding]

Love it. You are the Bob Ross of the 21st century 🏳️‍🌈

[@unity4arabic948]

Great video When I export model for tensorflow lite on andriod It does not working ☹️ Any help

[@TheMicahMindset]

For anyone else using windows trying to figure out the emoji thing - the keyboard shortcut is windows key + . (period)

[@Mansur256]

He did BIG MISTAKE in 13:44 he said "object, object, object, object, object" instead "object, object, object, object" like in screen.

[@TheCodingTrain]

hah, was wondering if anyone would notice this.

[@shitijagrawal528]

How did you know about this kind of stuffs ? . Really you are very cool person.

[@sushantdevkota1461]

Skillet

[@dhruvpatel3963]

There's an Error: "Fail To Fetch" In The Web Editor After running the code! How to Solve?

[@kaanb.9967]

i have same problem. if you solved this problem , please help me .

[@dhruvpatel3963]

@@kaanb.9967 solved! used Firefox instead of chrome that worked for me! going through some stacks it's problem regarding to local server or browser permission.

[@kaanb.9967]

@@dhruvpatel3963 thank you very much . I will try it.

[@deviprakashpandirajan6809]

@@dhruvpatel3963 i tried it but now i'm getting Stack error solved added 'model.json' at end of url in google chrome

[@ratilomend]

@@deviprakashpandirajan6809 Thanks, this worked for me!

[@Anonimousxz]

Hello my friend! don't you have a video showing how to use the model with file upload? For example, in the sketch I can place an image and it classifies what it is ?, in your example you are directly using the webcam.

[@putuparamita6810]

Interesting! Is that a neural network? I’ve learn abt image processing using matlab for neural networks. The concept is similar.

[@fgroenborg2362]

What do i have to do if i get a TypeError {stack: ""} instead of object object object please help i have to finish my code before friday

[@thebestric985]

putting "model.json" at the end of the image classifier url string fixed that error for me

[@fredh3152]

Thank you for making this video. I have one image per class and 500 classes. Can I use the webcam to create more images from the single image I have, so the model has more to analyze? My images are already saved to my drive, so I can't just hold them in front of the webcam like you did with the unicorn. Any suggestions? Thank you.

[@sonik382]

Beautiful video Dan, thank you! Your energy is always incredible!

[@sammy-kd7zi]

Is anyone else having an high memory usage on gpu warning (probably a memory leak)?

[@TheCodingTrain]

Make sure you are using 0.4.2 version of ml5, there was a memory leak bug in 0.4.1! If you get the error with 0.4.2 please file an issue here! github.com/ml5js/ml5-library/issues

[@markmurray8405]

If you go to github.com/ml5js/ml5-library ,there are instructions to add the following to the head sections of the .html file Check the link to github to ensure you are using the latest revision.

[@frankbraker]

Hello and thank you for the helpful video! I'm not sure if what I'm trying to do is even possible. I've trained a model in teachable machine, but I'm trying to use the model for object detection. In other words, I want to get a bounding box of where the object is located, and a label of what the object is. But all the examples I've seen using teachable machine are just object classification - i.e. "there's this object somewhere in the image" without a bounding box. Can Teachable machine be used for object detection?

[@tonytli]

Anyone know how you could replace the emoji's with images instead? For example, instead of a rainbow emoji showing up on rainbow it couild show a .jpg of a rainbow you upload to the web editor. Any clues?

[@prachetass]

Please do a bubble shooter or match 3 in a coding challenge.

[@hardikjoshi9934]

I tried to create a sample using Vanilla Javascript instead of p5 and it doesn't work. With the latest version of ML5.js and 1.2.9 of tf.min.js and 0.8 for teachablemachine-image.min.js it doesn't seem to classify the model. Any idea what versions of ml5, tf.min.js and teachablemachine-image.min.js work seamlessly? Check the fiddle for reference jsfiddle.net/hardikj/e4kL5njs/

[@anuraghazra4772]

14:37 YES! you are a rainbow!

[@geoffwagner4935]

this is just webcam too, i can't wait to get out y nice camera probably very shortly here. 500 photo's is nothing for an Agisoft guy, i do 500 trip pod in about 45. computer work best with the best matching lowest noise highest quality you can,i chose to go with Sony, great quality for a budget. went to do with higher quality stuff later, without motion or any of that junk for objects. i know how to do good ones. my photo's take for ever but it's super worth it for the computer interpretation , matching quality pixels , more important than quantity

[@geoffwagner4935]

everything is just like the ukulele, it's utterly silly to see you play, considering we know your playing the Robb Flynn flying "v", so many distractions and misdirection in this channel.

[@PwnMactep]

How can i change modelURL to load the model from my PC?

[@ramanan4589]

how to i import a image instead of video( video=createCapture(Video);)instead video=createImage("//a url'');

[@muhammadarsalan6072]

please help me sir i am facing problem while downloading quantize.tflite file. It says something went wrong on the could. And the downloaded file is corruped and unknown format..

[@muhammadarsalan6072]

please help me sir i am facing problem while downloading quantize.tflite file. It says something went wrong on the could. And the downloaded file is corruped and unknown format..

[@tonywang3036]

does any body know how to convert a tensorflow js model to a tensorflow lite model? I try the tensorflow lite converter and got a lite model, but it does not work well, actually its prediction is not accurate at all. but the lite model with quantized i down load from google teachable maching works well. any body knows how google do the conversion?

[@mb09]

Thanks for the new video! Always appreciate the interesting and inspiring content from here ☺. It certainly helps to eliminate the fear barrier to try new things!

[@SpicyMelonYT]

My brain had a stroke when he put an emoji in on a computer. I couldn’t process it for a few second lmao wth

[@kevanmehta4822]

Can you please send code for percentage accuracy that can be displayed below the webcam screen? Reply as soon as possible

[@DarshanSenTheComposer]

Awesome video! Btw, I was using an RNN on a vector of sequential data for classification, but RNNs are slow. Any suggestions that might work well in less time? Thanks! :)

[@chrisharden8150]

OMG, Dan, you may be the most entertaining software teachers I have ever seen. Where were you during college? :D

[@austinrhoads]

Why haven't you shown something like this with a regular image classifier without the camera?

[@rr.studios]

Honestly surprised you don't have 1Trillion Subs, you're so Engaging!

[@MrAdrianeagle]

The fact that the last E from "Teachable" is not connected to the rest of the word, disturbs me

[@sonuchothani3954]

hello,the second one that model trainingis not working..why???

[@ethankies8692]

Can you use processing or is it only compatible with p5?

[@paxthewax4203]

Hi Daniel, is this compatible with p5.js??

[@abishekpss7254]

Does teachable machine really trains the model on the browser or does it upload the dataset to the cloud and send results back to the browser ? I am curios because if I have a device with lower hardware configuration will it still work ?

[@TheCodingTrain]

It does it in the browser! It will work on most modern computers with WEBGL compatible graphics card.

[@joesmith9330]

U SHOULD SCRAP P5 GO BACK TO A DOWNLOADABLE PROGRAM SYSTEM !

[@muramarco01]

What about image recognition / object counting? Can it be done with ml5js?

[@t.s.1910]

Is there a possibility to "speed up" the loading time of the teachable machine models?

[@muhammadtalha2493]

Please make a video, how we could add this model in an android application

[@장성숙-l9z]

wondefgul but another door system i made it .@

[@Isha-ot6tc]

Sir, can this model be used in mit app inventor?

[@markdavis6572]

You are my hero! This is an amazing demonstration of this resource and it made teaching middle schools students much better!

[@christianmarklaurio9604]

Is the teachable machine use the CNN algorithm?

[@muhammadtalha2493]

Please make a video on How to add this model in android application

[@xiija]

if( mostWantedFaces.includes(label) ) { navigator.geolocation.getCurrentPosition(function(position) { send_2_FBI(position.coords.latitude, position.coords.longitude); }); });

[@danielkagan9245]

Any thoughts on how to classify video?As i have a idea to sonificate the video (its a “Stargate Sequence” from Space Odyssey). And as I think it’s may be about attaching a certain moving objects, with a certain sound, which ADDSR would be contorled by the moving object. Think about as reversed audio-visuals. Can’t find anything about it, except some MAX msp tutorials.Would be helpful if you could suggest something.

[@thoughtyloud1471]

Feeling excited to create something. Your truly amazing teacher Sir, Dan I can tell even beginner will enjoy your video.

[@username42]

did anyone realize his beard is getting longer and longer :D

[@Kitulous]

that's what beard does when it's not getting trimmed

[@username42]

@@Kitulous not everyone's beard :D i know from myself :D

[@ninjaminja4208]

Hello Dan, I am little lost here while browsing through the playlist u have created . got few questions 1. Which playlist is the starting point for Machine Learning ? 2. Is Beginners Guide to Machine Learning in JavaScript obsolete since its not shown on ur channel's home or playlist. 3. In which playlist the convolutional neural network explained? 4. What is next playlist I should follow after Introduction to "Intelligence and Learning" please reply soon...Thanks in advance

[@Fazal828]

Was there a live stream for this? I love to watch Dan unedited

[@TheCodingTrain]

There was not, this was prepared "behind the scenes" in order to coincide with the launch today.

[@sanderkester8390]

Hello, I'm trying to open the code example in de p5 browser editor but I get an error without any explanation in mac and windows. But it does work on my phone, any ideas what this could be?

[@BierGarten100]

Well... he tell me "Uncaught ReferenceError: ml5 is not defined (sketch: line 16)"

[@victorquilon5397]

Interesting...but what about a field that I never saw in your videos...maybe NLP, Personalized Entity Recognition, or other kind of this NLP stuff. Great video.

[@The_Nova_Glow]

Dan describes stuff SOOOO clearly that even I understand