BIM geometries converter

Jon Lopez de Guereña — Wed, 07 Dec 2022 06:43:44 +0000

I made a wrapper around Ifc Open shell, a library that allows you to convert IFC files to other formats (dae, obj, stl, etc) using different options: github

It's built on React, Nodejs, three.js and IfcOpenshell.
The backend can be deployed on his own, which can be very useful not only for the web (where there are very good solutions such as ifc.js) but also for unity or unreal applications where plugins for this type of task are quite expensive.
The front allows you to upload and preview your Ifc in a webgl canvas before downloading it.

It's fully dockerized(into three different containers) and can be deployed anywhere using pm2 to manage the nodejs process, allowing you to work in cluster mode for better perfomance.

I probably would have done better and faster with gpt3 but I wouldn't have enjoyed the process as much :D

Give it a try:
ifc converter

I built a website image scraper + AI classifier

Jon Lopez de Guereña — Fri, 21 Oct 2022 07:56:31 +0000

I have been a member of the community in the shadows for many years, and I felt like it was time to write a post 😊.

As python is one of the vehicular languages in artificial intelligence concerns and it's incredible how AI and ML are improving day by day, I wanted to start learning about this subject.

At first I thought to start with one of the many python learning websites, but I dont feel like it's the best way to start learning a new lenguage when you already know how to code (just a personal opinion which works for myself).
I prefer to set some goals and then look for the information.⭐

So I decided to build a tiny pet project to get contextual information about a website based on it's images so i could struggle with python and AI in a more realistic way.

The application is able to scrap images from any website in order to classify them with tensorflow.
It's built on FastApi to expose a rest api for an easier management, and it's fully dockerized to deploy it anywhere (and not having to fight with CUDA drivers).

I think it could be useful to create image datasets or just analize websites to get contextual information about them.

Of course any suggestions are welcome! and feel free to check it out and give it a try 💗

jonlo / py-web-image-scraper-classifier

py-web-image-scrapper-classifier

Description

This application, exposes an API to scrap and classify images from a given URL.

It's fully dockerized so you can deploy anywhere.

You can easily select the tensorflow model to use for the classification by changing the model_name variable in the imageclassificator.py file.

The options are:

/scrap To scrap the images from the given URL

/classify To scrap and classify the images from the given URL

/image To get the image from the given URL

Uses fastapi, tensorflow and uvicorn.

Usage

If you have CUDA installed, you can just run uvicorn

uvicorn main:app --reload

go to http://localhost:8000/docs

Docker

If you don't have CUDA installed, you can use the docker image

docker build -t py-web-image-scrapper-classifier .
docker run -p 8000:8000 py-web-image-scrapper-classifier

View on GitHub