Case studies

We are happy to demonstrate working objects recognition in action:

Upon the request of an innovative aero-vision company, Provision Lab developed a vision application that tracks and identifies objects in real-time video. Our technical team faced a challenge to build a lightweight but powerful application that can count the number of cattle and sheep, on the basis of the video attained from a drone. There were options both for video which recorded on MicroSD/SD card as well as through real-time video streaming. We managed to quickly cope with this technical challenge by using an objects detection algorithm, also applying KSF tracker and neural network training.
Technologies: c ++ / cafee / Qt

We are happy to demonstrate working objects recognition in action:

A next generation Internet gaming/gambling company asked our team to develop an innovation vision application that can track and identify objects in real time. Our task was to create an application that detects and recognizes cards in the casino on the basis of the video attained in streaming video from the security cameras. This technical challenge was resolved by the automated object detection algorithm, also applying convolutional neural network training.
Technologies: c ++ / cafee / Qt

We are happy to demonstrate working traker/detection recognition in action:

An early stage movie startup company addressed Provision Lab with request to develope a vision application that tracks objects in real-time video. Our developers successfully coped with the task to build an app for AF Filming equipment on a basis of video attained from the camera. In addition to autofocus issue, this technical problem was resolved with an Object retention after passing through barriers and face recognition. Provision Lab successfully coped with this task by using detection of objects and applying KSF + Yolo tracker and neural network training.
Technologies: с ++ / cafee / Qt / ksf / yolo

We are happy to demonstrate working face recognition in action:

A very promising security systems startup contacted Provision Lab to develop a vision application that recognizes and tracks objects in real-time while the live video stream is being broadcast from a device or camera. Our technical team had to build an application for human object and human motion recognition on the basis of video received from the camera in real-time. This technical problem was solved using detection of objects and applying opencv tracker and neural network training.
Technologies: с++/cafee/opencv/Qt

Detection of broken threads:

At a factory for tailoring T-shirts, you need to quickly detect the breaking of the thread on the looms. In the past, control of breaks was carried out by employees who were supposed to control about a hundred machines and more than 500 threads, which was economically expensive. Our programmers developed a program that, with the help of a camera, allows you to recognize a broken thread with a probability of 95%. It will help to decrease downtime in factories
Technologies: c++,opencv,tensorflow

Panoramic view:

We wanted to introduce a part of the Project- 360° We made it for the military needs of the country, a real-time video shooting system was developed in the 360 ° video demo version of the alpha testing system. Additional data can be provided on request.
Technologies: c++,opencv,tensorflow,Qt

Detection of dangerous objects:

Identification of dangerous goods at airports is a key task of the airport security service. At the current stage, millions of dollars are spent on these tasks by each airport. ProvisionLab is pleased to participate in the project of our Israeli partners in the detection of dangerous objects. 3D system for reconstruction and processing of data from various sources (scanners, cameras, thermal cameras) and interface interaction with the user. This solution allows the creation of a 3D reconstruction of the object and the formation of its model for further analysis.
Technologies: python, PyQt,tensorflow,dlib

SG intrusion detector:

This is a service, which allows detecting people in the specified area. This is provides an easy way to use the detection of people for ordinary users who want to automate this process. This service is quite easy and convenient to use to notify users that in some zone there is a person, for example, an unidentified person on a protected object, penetration into the house and so on.
Technologies: c++,golang,opencv,tensorflow,scyllaDB