Artificial Intelligence Based Computer Vision Systems

We Offer High Performance Solutions That Will Meet Operational Needs

As the use of cameras increases in every field, it becomes even more important that captured visuals are examined quickly and reliably. We develop technology which excels in this field and transform this technology into operational solutions.

Specialized for various needs in different fields, we develop object & face recognition systems based on deep learning models. With their expertise on single & two staged real-time object detection, our computer vision team has produced several modular solutions to detect objects on image, video and stream inputs. Our solutions can easily be integrated with existing systems.

Real Time Video Processing

We are developing scalable systems where deep-learning-based analysis such as object detection, super-resolution and anomaly detection is performed on any type of camera feed, where results are displayed to the end-user and certain alarms can be set.

Object Detection in Image and Video

We offer deep-learning based high precision & real time object detection solutions for image and video. We can customize our pipelines for specific needs to achieve state-of-the-art results.

Relating Images, Searching Images, Finding Source of an Image

We offer our customers the opportunity to benefit from finding the similarities between images through deep learning technologies.

It is possible to benefit from our solutions in many different ways such as spotting similar images within a large image set, detecting product/brand images and clustering specific images.







Image and Video Summarization

We offer a fully automated summarization solution that extracts semantic information from visual data and describes the scene in natural language. While these interpretations support decision-making mechanisms, they also allow the storage of visual data in more specific structures.

Face Detection and Recognition in Picture and Video

With the most up-to-date deep learning algorithms, we detect faces in video and pictures in real time. We create tailor-made solutions by optimizing our detection pipelines for specific needs.

Super Resolution for Image and Video

In a given image or video input, we enlarge the resolution of a desired region or the whole scene by using deep learning based super-resolution models. In this way, we improve the image and video inputs obtained via distance-shooting or shot in bad visibility conditions. Our super resolution system can be used for pre-processing purposes ahead of object detection, or it can be used as an end-to-end system.

With the deep learning-based algorithms we have developed, we are generating super resolution models for different video conditions. We provide end-to-end service for our customers from data set preparation to system installation.

Fog Removal for Image and Video

We develop models that work in real time with deep learning based algorithms on RGB cameras. We solve the problem of removing fog from scenes completely in software without the need for extra sensors of the camera.


Detection of Unusual Events like Anomaly, Panic, Attack etc. in Scenes and Raise of Alarm

We develop real-time activity detection models for stationary and moving cameras. In this context; we put forward systems that find categories such as group activities, banners, protests, police, flags and raise alarms if necessary. We produce end-to-end solutions for data set preparation, model development and model training for different categories.

Generating Synthetic Data on Simulators

We provide synthetic video-data generation services for different needs on different simulators. Based on customer needs, video-data production can be configured to contain specific scenes at specific weather conditions. Produced video-data can also be configured to include label formats with specific characteristics.

Solutions Compatible with Images from Different Camera Types (IR, RGB, etc.)

We have real-time object detection, face detection and recognition, superior resolution solutions compatible with different camera types (IR, RGB, etc.). The algorithms we develop can self-recognize the difference between IR and RGB scenes and produce more accurate results by optimizing the deep learning model.