ml5js / ml5-adjacent Goto Github PK

Here's a list of the interactive embeds currently referenced in the article with what is here so far in the repo (thanks again @AidanNelson!)

• - drag and drog image classification (can use ml5-homepage-demo)
• - webcam classification
• - Pitch Detection
• - Style Transfer
• - word2vec
• - PoseNet
• - LSTM generation

We can file individual issues for things we need for each one, this will just keep track of the full list.

@oveddan can you take a look at this sketch? (It's for an embed demo in the adjacent article).

https://github.com/ml5js/ml5-adjacent/tree/master/07_PoseNet

I'm trying to get it to work with a sample starter image then the user can press a button to start detecting poses from the video. Right now it works but I'm not getting any results for the starter image (which is just a placeholder, so maybe a bad image?)

We may need to incorporate loading animations if any are slow to load for the Adjacent page.

See: https://twitter.com/jaymesdec/status/1217158118859247622

For any of these that involve a video we should have the reader "opt in" to starting the video.

@stephkoltun do you think it makes sense to do a separate image embed to the side or perhaps what makes sense is to have the page load with a static image demonstrating the idea and then a button to start making it interactive?

Thank you @AidanNelson for getting this repo started, hooray!

I'm not sure to what end we will want these to appear very generic or match the ml5 design or the Adjacent design. @matamalaortiz perhaps if you have some time to do a design pass over these along the lines of the homepage embed?

Or feel free to just provide some generic CSS specs and I can try to incorporate?

@cvalenzuela can you take a look at:
https://github.com/ml5js/ml5-adjacent/tree/master/02_ImageClassification_Video

I couldn't get the modelReady() callback to work so just used setTimeout(). Also it didn't work with loadImage() but that's not a big deal for this. I'm rushing through these so easily could be missing stuff, doing something weird.