Azoft RnDCase Studies

Research and Development

Azoft R&D department handles the most challenging, non-standard issues that arise when implementing software solutions. Many of the problems tackled by our R&D team strongly rely on scientific research and may involve technical risks. If you have an idea or a technologically complex project but aren't sure how to make it work, Azoft R&D team can study the problem and come up with a technologically sound approach. In this section you can see some of the projects we have recently completed.

Classification of EEG Signals for Brain-Computer Interface

By Ozhiganov Ivan on October 20, 2015

Classification of EEG Signals for Brain-Computer Interface

Using brain-computer interface could facilitate everyday life of paralyzed or disabled patients. With a spate of interest in this topic, Azoft R&D department, together with Sergey Alyamkin and Expasoft participated in "Grasp-and-Lift EEG Detection" competition organized by Kaggle. The competition was dedicated to classifying various movements of the right hand via EEG (electroencephalogram) for a brain-computer interface development.

Machine Learning Methods in Video Annotation

By Ivan Ozhiganov on July 10, 2015

Machine Learning Methods in Video Annotation

Research and development in the field of image and video processing is an important and popular task today. In the era of mobile technologies almost anyone can make a video and upload it on youtube or any other hosting service. To find relevant videos, they need to be classified. One way to handle video classification is by using tags annotation.

Developing a Face Recognition System Using Convolutional Neural Network

By Ivan Ozhiganov on May 14, 2015

Developing Face Recognition System Using Convolutional Neural Network

The last decade has become the breakthrough in artificial neural networks development : this technology now is widely used in the areas, where traditional algorithms could not cope with the tasks. Spam filtering, contextual advertising, gambling, recognition of texts and characters, and more — neural networks show great results.

Security is one of the most important applications of neural networks. Networks are used for surveillance and monitor suspicious banking transactions. Azoft R&D department is currently engaged in the development and training of such a network: we create a face recognition system that is invariant to face positions in the space, turns, lighting and facial expressions.

Real-Time Video Processing Algorithm for Instant License Plate Recognition in iOS Apps (using OpenCV & GPGPU)

By Ivan Ozhiganov on May 29, 2014
Real-time Video Processing Algorithm for Instant License Plate Recognition in iOS Apps (using GPGPU & OpenCV)

Since Azoft’s first OCR project we were exploring the GPGPU technology and the opportunities it provides for character recognition. Now our team is working on an iOS app prototype for recognizing license plate that uses GPGPU. The key feature that makes our solution unique is real-time video processing that makes instant number recognition possible.

NFC Alternative: Transferring Data Between Mobile Devices Using Ultrasound

By Ivan Ozhiganov on October 9, 2013
Data Transfer Between Mobile Devices via Ultrasound Azoft is working on an innovative short-range wireless technology that could serve as an excellent alternative to Near Field Communication (NFC). The idea is to transfer data from one smartphone to another using ultrasound. This technology looks very promising and could be applied in many ways, especially in the area of mobile payments. Our ultrasound technology has a big advantage over NFC: it does not require any additional hardware (i.e. NFC chips) and could be compatible with virtually all smartphones on the market today.

Applied Use of M2M-Technology: Road Sign Recognition in iOS Apps

By Ivan Ozhiganov on July 9, 2013

While machine-to-machine (M2M) technology isn’t a new concept, the area of in-car apps is growing rapidly and providing new and exciting business opportunities, particularly for mobile developers.

Azoft’s first M2M project focuses on integration of mobile apps with car in-dash system. Our team is currently working on an iOS app prototype for recognizing road signs. It's considered to be the first step to a so-called smart dashcam, driver's reliable partner on a road.

Optical Recognition of Credit Card Numbers in an iOS Application

By Ozhiganov Ivan on May 30, 2013
Our recent research on credit card number recognition using an iPhone camera consists of four stages: card image capturing, text area detection, splitting the detected text into four blocks with digits, and finally the digits recognition.

We use Sobel operator and Hough Line Transform method during the image capturing stage and the method of strokes in the splitting stage. The obtained 16 areas with digits are transferred to a multilayer convolutional neural network which contains 3,500 images in the learning base. Even such a small learning base allows our prototype app to provide quite accurate results when recognizing credit card numbers: just 2 errors per 100 test samples.

Algorithm for Identifying Barely Legible or Embossed Text in an Image

By Ivan Ozhiganov on April 11, 2013
The challenge of detecting embossed text, like that on credit cards, has been occupying minds of our R&D team for quite a while. Even though all traditional approaches show little or zero results, Azoft specialists are making small steps closer to the solution. The most recent step made is a modification of stroke width algorithm the implementation of which showed quite well, not yet excellent, results in identifying embossed symbols.

Grayscale Image Sobel Operator Method Otsu Method Strokes
  • Pages:
  • 1
  • 2
  • 3

Request a Free Quote

Please enter the result and submit the form