Computer vision

Object recognition applications

Nowadays, computer vision is trending technology. It is highly in demand in the security and surveillance industry, self-driven cars, and entertainment apps to name a few.

This surge in popularity of computer vision is largely due to the emergence of state-of-the-art deep learning technologies that are able to solve computer vision tasks with very high accuracy, something which was considered unachievable a decade ago.

In this post, we will see the most popular and creative computer vision applications.

Image Classification and Localization

Image classification involves assigning a class label to an image, whereas object localization involves drawing a bounding box around one or more objects in an image.

With AI.dielmo, you can also choose between different parameters for the classification and localization of your images and videos:

Confidence threshold
Blur factor
Number of threads
Packet processing size
Stroke size
Buffer width
Buffer height
Merge type
Merge intersection
Laplacian threshold
Process by parts
Window Width
Window Height
Overlap Windows

Object detection

This application of computer vision is a more generalized version of the above task(Image classification and localization). In this task, an image may contain more than one object which needs to be classified and localized individually. Self-driving car technology fundamentally relies on object detection to navigate through the roads.

AI.dielmo can be used to check if a photo has a certain element or not, to make a quality control, or to crop the element(s) in a new image, etc.

See how object detection works with AI.dielmo.

Semantic and Instance Segmentation

We all know that image is a collection of different pixels. The aim of segmentation is to group together the pixels that have similar attributes. There are two types of segmentation – 1) Semantic Segmentation and 2) Instance Segmentation

In Semantic Segmentation, each and every pixel is classified into a class label. In simple words, semantic segmentation does a pixel-wise classification i.e label each pixel with a class.

Instance Segmentation is the combination of two task object detection and Semantic Segmentation. i.e. Instance Segmentation=Object detection+Semantic segmentation.

For example, in AI.dielmo, we use it to vectorize on orthophotos buildings, swimming pools, etc.

Object Tracking

Object tracking is the process of locating a moving object (or multiple objects) over time. Visual object tracking is an important element of many applications such as person-following robots, self-driving cars or surveillance cameras, etc.

Using AI.dielmo we can, for example, customize an algorithm that, from a high resolution video, extracts the frames that correspond only to the electrical towers and selects the best of them for each tower, thus increasing the productivity of the inspection works.
If you need a customized algorithm for your project, in AI.dielmo we can do it.

Image blurring

The objective of this task is to blur objects in an image.

With AI.dielmo, you can also choose between different parameters for the blurring, such as:

Confidence threshold
Blur factor
Number of threads
Packet processing size
Stroke size
Buffer width
Buffer height
Merge type
Merge intersection
Laplacian threshold
Process by parts
Window Width
Window Height
Overlap Windows

In this post, we saw plenty of computer vision applications right from the basic ones to the more advanced ones. Find the one that best suits your needs for your project.