Computer Vision

Because 1000 pairs of eyes see more than 1.


“Computer vision is an interdisciplinary scientific field that deals with how computers can be made to gain high-level understanding from digital images or videos.
According to Wikipedia

The purpose of computer vision is to make the computer understand the story that an image tells. This can relate to both photos and (live) video. For us humans, this is a simple thing. We see (usually) at a glance what an image wants to tell us. It's a different story for computers. It is even a very complex task for that.

What exactly is computer vision?

For example, when we see a book on a photo, we see that it is a book, what kind of book (large, small, hard or soft cover), the title, the author, the genre, how it is part of the photo (subject, is somewhere in the cupboard, is on the table, is used as a door stopper) and so on. A computer only "sees" the pixels, or rather, the intensity of the pixels of the photo. For a computer there is no direct link between the pixels, let alone that a computer can give content to what is on an image.

Computer vision is used to unravel the content of an image.

With computer vision, basic edits are used on the image to be investigated, image classification and deep learning to give content to the story in the photo.

Basic operations

Basic operations of computer vision include things like reading and displaying an image, drawing lines, rectangles, circles and text, converting to different color schemes, resizing, cropping, rotating, masking, morphing, smoothing, blurring, determining color gradient and edge and contour detection.

Basically the things you can do with photoshop, but without the fancy UI. Special libraries such as OpenCV are used for this.

All these techniques are used in combination with each other in image classification and deep learning scenarios.

Image Classification

Image classification is the assignment of a label from a predefined set of labels to an image.

To make the classification easier, the images are edited to remove unwanted elements from the image, or to make the subject to be classified more detectable.

Machine learning and deep learn are used for classification and deep learning algorithms.

Deep Learning for Computer Vision

Deep learning for computer vision includes the use of neural networks to understand the content of the image.

Training and evaluating a network is a tough task that must be done just right enough. If a network is not properly trained, it will not be able to correctly classify an offered image.

Examples of Computer Vison

Computer vision is used in many areas. Think of license plate recognition, face recognition, quality control (deviations from the reference), medical applications (detection of fractures, joint wear or tumors) and, also hot today, operating systems for self-driving cars.

Deep learning is often not required to detect an object in itself. Deep learning is usually used to identify the object.

Want to know more?
I would like to discuss this with you.
Send me an email (here) or
send me a DM via my LinkedIn profile below.


Peter Lans

Development Consultant