computer vision


We leverage images and videos to power the next generation of products and solutions.

Pento customer - CAF Banco de Desarrollo de América Latina
Pento customer - BASF
Pento customer - Duke University
Pento customer - BunkerDB

What we do

Large collections of images contain a lot of intrinsic value. Understanding them automatically will enable you to improve several different processes in your company.
Recognizing what’s present on your images can help you in a number of different ways: reduce the time spent on costly manual image labeling, enhance your users’ search experience, or improve your data by removing duplicates or out-of-distribution images.
We help you understand what can be extracted from your images and what processes can be improved with this information. With a clear goal in mind, we build this system and advice you on how to integrate it with your existing workflow, as frictionless as possible.

Visual analysis

Hundreds of modifications stand between a raw photograph and a ready-to-be-published final image.
Retouch your photos automatically, removing and changing their background, adjusting the color, lighting, and shadows. By performing these manipulations without manual intervention, benefit from improved speeds and reduced costs, allowing you to scale up as much as your business needs to.
For an out-of-the-box solution, check out SolidGrids , our plug-and-play solution for e-commerce, or reach out to us for custom-made solutions for your requirements.

Image manipulation

Analyze videos to detect interesting objects, understand people’s movement, catch product defects or anomalies, and more.
Our solutions allow you to automatically analyze hours of video, unlocking opportunities that would be not possible if they are done manually, or would take just too much time.
These systems can be applied to solve a variety of challenges, from detecting defects in your products to checking personnel security, such as detecting helmet usage. Feel free to contact us so we can help you determine if it is applicable to your case.

Video analytics

This is how we approach projects

Project approach step - Identify opportunities

Identify opportunities

Starting strong in a project can determine its success in the future. The key is to begin by identifying opportunities with true value. A clear goal with a significant, measurable impact is a must and helps every party involved in the project. Thus, we work alongside our customers to identify or adapt an already existing idea when necessary, focusing on what you probably really care about: improving KPIs and reaching business goals.
Project approach step - Research


Once the project is clearly scoped, we evaluate different available alternatives to tackle it: from classic techniques to state-of-the-art solutions. More often than not, ML projects are experimental by nature and carry some uncertainty and in this way, we reduce their risk. Throughout this phase, we work on proofs of concept and gather preliminary results to assess its feasibility.
Project approach step - Development


Backed by our learnings in our previous step, at this point we have a clear idea of the problem’s domain and the necessary steps to address it. We start by getting baseline results to compare to. We iterate on the solution we build, incrementally improve upon it. Once we reach the intended results, we get our models out on production, where they actually shine and deliver value.
Project approach steps - background