Mickaël A.

Hi Simon. Thanks for taking time to review this. Yes, you are right. The precision is lost, and in case of billions of images it's better to increase the number...

Hi. I added a speed test with a "flatint" driver where words are stored as an array of lon ints. It is not as fast as the text array. From...

> @miqwit are you still pursuing this? Also have you tried testing on a very high number of images? I would be curious to see a graph of the stats...

Yes, well, I won't test it on that amount :) I'll see if I can raise it to a more significant number. At least I'll pull some statistical data.

Hi @taylorjdawson. I improved my script by generating graphs (with matplotlib) about various performances. I am in the process of validating the concept with some probabilistic approach (this should answer...

You can find generated plots on my machine [here](https://github.com/miqwit/image-match/tree/words-flat/tests/plots) Here is the query time one:  This needs more explanation about all of them, I plan to do this soon...

Hello @heipei. No, not really... This would need to dive in the ES engine. It is based on Lucene and open source, so I guess it's feasible, but I don't...

Hello. +1 here. Any progress on this? Would you need help to implement it? I have this code in PHP in my project (reading a geojson file), which does the...

I updated my algorithm, as it had a problem. I will be happy to contribute to your library, I'll try to find time. On the other hand, it would be...

I worked on the bounds topic this morning, before seeing there is some work done in `develop` :( Still, I'd like to submit my code in a PR for a...