Mattia

I sped up the execution of a bash command line >500x to make it a breeze to usePhoto by Chris Liverani on UnsplashWhen you have to process data in textual or tabular forms, one of the long-time favorites is surely GNU Bash, the Linux flagship shell with “batteries included”. If you have never used it, you are missing out a lot and should definitely give it a try.The tools coming with Bash follow the Unix philosophy of “do only one thing and do it well”, and are super-optimized for many different tasks. Find, grep, sed,…

From a digital elevation model to a 3D meshPhoto by Planet Volumes on Unsplash3D meshes can be used to represent geographic data such as terrain, buildings and other structures. This mesh can be useful for a variety of purposes, such as urban planning, environmental analysis or virtual reality simulations. However, the process of creating a 3D mesh of a geographical area is not always straightforward, but this guide covers all the necessary steps.To generate a 3D mesh of a given area, elevation data for that area is…

Build and train a segmentation model with a few lines of codeMartinThoma, CC0, via Wikimedia Commons (edited)Neural network models have proven to be highly effective in solving segmentation problems, achieving state-of-the-art accuracy. They have led to significant improvements in various applications, including medical image analysis, autonomous driving, robotics, satellite imagery, video surveillance, and much more. However, building these models usually takes a long time, but after reading this guide you will be able…

A foundation for current and future global challengesPhoto by NASA on UnsplashIn a previous article, I discussed in detail everything you need to know about working with geospatial raster data. This type of data is associated with a specific geographical area. However, it’s not possible to study how that area has changed over time by analysing a single raster file. This guide aims to introduce the reader to satellite image time series and explain why this data is incredibly important in addressing current and future…

Bring joy to your fellow developers by making your software easy to use through environment and build automation. With code examples in Python and Hatch.Photo by Carol Jeng on UnsplashMost developers hate legacy software, why? “Legacy” in our industry means a codebase that has been in service for many years, usually the original developers are no longer in the company, and nobody is able to really maintain it.Some important ingredients in the legacy software recipe are: lack of documentation, hard-to-understand code,…

Let’s dive deep into signal representation, Fourier transform, spectra and harmonics.Photo by Richard Horvath on UnsplashThis article is part of a series:IntroductionBefore going deeper into the speech enhancement rabbit hole, we really need to make clear some concepts about digital signal processing. In the previous chapter of this series, we introduced some concepts and added links to further readings.In this second part, we are going to explore how digital signals are represented, how to change representation, and why…

An introduction into the concepts, methods, and algorithms that allow us to improve the quality of degraded speech or suppress noisePhoto by Wan San Yip on UnsplashSpeech enhancement is a set of methods and techniques aiming at improving speech quality in terms of intelligibility and/or perceptive quality using speech audio signal processing techniques . It has many practical use cases, including cleaning speech signal from noise in hearing aids, or recovering a hard-to-understand speech signal from a noisy…

Your guide to understanding how deep learning and pre-trained models can be used for evaluating automatic voicesPhoto by Vika Strawberrika on UnsplashSpeech synthesis, or text to speech (TTS), is a decades-old technology that came back strongly in the last years thanks to the huge improvements provided by deep learning.Synthesized voices sound more and more natural over time, and it becomes harder and harder to distinguish them from human voices.This is the general trend, but still different systems reach different…

How to automate before having a production-ready MLOps platformPhoto by Campaign Creators on UnsplashWhen working in a cluster or any other kind of shared processing system, one problem that emerges soon is how to share easily data processing tools. All users have their own environments, libraries, and the functioning of software can secretly depend on some paths defined in the .bashrc. It is a pity to write some valuable data processing tools if then it is not possible to share it with the people that really need…

Combine Deep Learning with 3D data processing to generate a meshPhoto by Alvaro Pinot on UnsplashGenerating a 3D mesh from a single 2D image seemed a very hard task some years ago. Nowadays, thanks to the advancement in Deep Learning, multiple monocular depth estimation models have been developed and they can provide a precise depth map from any image. Through this map, it’s possible to generate a mesh by performing surface reconstruction.IntroductionMonocular depth estimation is the task of estimating the depth value…