Object detection

Object detection using neural networks is one of the key technologies that finds its application in various fields, from the microcosm to industry and everyday life.

We are surrounded by objects that artificial intelligence can detect.

In medicine, for example, neural networks can be used to automatically diagnose diseases based on medical images. For example, a neural network can detect tumors on x-rays or on magnetic resonance imaging.

In the automotive industry, neural networks can be used to detect and recognize road signs, traffic lights, other vehicles, and pedestrians on roads. This can greatly improve road safety as well as autonomous driving.

In banking, object detection using neural networks can be used to protect against fraudulent transactions, as well as to automatically recognize the authenticity of customer signatures.


Object detection technology using neural networks is based on training a neural network to recognize objects in images. This requires a large amount of labeled data on which to train a neural network. We use hundreds, thousands, or tens of thousands of images to train a neural network to distinguish between objects.
After training, the neural network can detect objects in new images, even if they differ from those on which training was carried out.

In general, object detection using neural networks has great potential in many areas and can significantly improve the quality and safety of people's lives.

As an example, we can cite the case of a search engine implemented by us based on neural networks

The team was tasked with implementing an API with the following functionality:
API user management (create, edit, delete)
Image storage management (create, edit, delete)
Image management (creation, editing, deletion). Saving images to  defined storage.
Search for images with the following user-defined parameters:
Number of similar images
Maximum distance between searched images
Search for images with detected objects (additional neural networks were used). This functionality allows you to search not just the entire image, but and only the main object (the central or the largest in area), or all objects found by additional neural networks. Also, the  API has the functionality to determine the maximum number of possible detected objects on the  image.

An additional functionality is the ability to search for an object without a background, the object is also separated using a separate neural network.

Trash detection