spousty

Oh so did you run the CNN on pre-classified data

It's due to the fact that they used satellite imagery to create the embeddings. The map is just for visualization. They probably used 5 or more bands of the satellite data which means each pixel is going to be slightly different due to things like depth, amount of silt in the water, amount of plankton....

Having worked on these types of problems before the model is doing a pretty great job matching pixels.

How did you generate the embeddings. The vectors are relatively small for all the embedding I have seen built from image and nlp models.

Which copernicus bands were you using? Did you augment the data with DEM info?

Is there is a Github repo for this?

Should be fixed now - thanks!

Thanks for the this great information. In the hands on tutorial linked in the article I address some of these constraints you talk about.

Thanks for the feedback. I think the article was really trying to whet people's appetite for the hands on tutorial linked in the article. I wrote the blog post just to give people enough to: 1. Know they can actually store, query, and update JSON in PostgreSQL 2. Get curious and dig in deeper with the free hands on training.

I guess PostgreSQL doesn't do everything for me :D Thanks for catching those and I will be updating the post soon.

Did you get a chance to try the hands on exercise?