Computer Vision

Just like human vision, computer vision processes visual information from images, videos and live data.

How can you apply computer vision solutions?

Digitizing printed and handwritten documents

When text is written backwards, hidden within complex shapes, or obscured by pictures, the individual characters can be difficult to recognise — a process that is called Optical Character Recognition (OCR).

Before OCR can be used to extract the text, other methods of computer vision are used to clearly define the area in which the text is located. In each individual case, various different font and language improvements might also be needed to ensure the most accurate character recognition.

Our team have deep expertise in OCR, and have implemented the technology in various projects. For instance, with Chart Recorder, we developed an algorithm that enables data to be captured from gas flow charts. In other words, it allows recognizing each trace on the chart and generating an output table with the needed metrics of pressure, temperature, and water.

This same technique can be applied to many other spheres: For banking, it’s possible to build a solution which allows data from paper documents to be transmitted to a database for further processing. For retail, a store selling alcohol could implement a software tool capable of processing the date of birth on ID cards.

Identifying objects on photos and videos

For object identification, two technologies can be applied — machine learning and computer vision. Simple tasks might be approached with computer vision, but more complex objects might require machine learning, neural networks, or a combination of the technologies.

Solutions can also be developed that detect certain events from video streaming. For example, the state of construction sites and industrial facilities could be compared to the optimal conditions using computer vision.

In one of our projects, we implemented an algorithm that can count falling drops of liquid from video or live data. This can be used in industries like high-precision manufacturing to forecast equipment breakdown, or in pharmaceuticals to ensure accurate dosing.

In the E-commerce sphere, object identification can be used to identify products from photos, and search for similar products.

Defining the sizes, shapes, and positions of objects

Identifying the exact size and shape of an item can enable the perfect placement of an object in a warehouse, optimizing the storage space and increasing capacity. Solutions can also be developed that allow you to check whether or not items are located correctly.

Controlling product quality

Computer vision allows quality control to be executed efficiently in a range of environments. Completed products can be automatically inspected for the presence of defects, the quality of packaging can be ensured), and components can be checked for the right fit.

One of our recent solutions makes it possible to ensure that the right volume of liquid is in a bottle, and that the label is placed correctly.

Tracking the movement of objects

This can be useful for navigation systems, allowing autonomous vehicles to detect, and monitor the movements of, other moving objects.

A combination of two or more tasks

Sometimes projects require a combination of the above tasks. Our team are able to provide integrated, multi-functional solutions. One of such projects is “License plate recognition system”. First, it recognizes the license plate number of a vehicle. Second, it sends a signal to the parking lot gates to be opened automatically.

How do we approach Computer Vision projects?

Across all our projects, we use the languages C++, Python, and Java.

Does your idea need computer vision expertise?
Our team are ready to handle the challenge!

Optical Character Recognition
Optical Character Recognition
Full cycle of software development
Full cycle of software development