DEV Community

Cover image for Background removal on car images - Machine learning project
Ibra
Ibra

Posted on

Background removal on car images - Machine learning project

Introduction

The goal of this project was to have an algorithm capable of removing the background of a vehicle image automatically with machine learning.

Technical

This was done with Python using the framework Keras, Tensorflow and a number of known python libraries like Numpy, Pandas, OpenCV, etc…

The algorithm uses a lot of CPU power to predict and remove the background and by consequence, this raises a scalability issue, it can get very costly to deploy and scale something like this.
So to get something high-performance and scalable, i did a lot of work optimizing the code, packaging, removing useless code all to reduce the size of the end result.

After weeks of hard work, i was able to deploy it in a Serverless environment on Google cloud with Cloud Functions.
So now, i had a ML model that is fast, scalable and almost totally free (1mil operations per month for free).

Source:
https://www.ibragim.fr/portfolio/detourage-automatique-voitures-ia/

Top comments (0)