Logo icon

Study of possible automatic detection of tackle in soccer

Developed for: Intel Inc., Nizhny Novgorod, Russia

Purpose: study of possible detection of complex sports technical tricks by the example of tackle in soccer

The main objective of the project is to research possible automatic detection of tackle in soccer.

As a part of the study, we have conducted a comparative analysis of the existing solutions to automatic initial data collecting and processing for sports analytics, assessed the required number, location and type of cameras; considered the possibility to apply to the subject area and determined the direction of enhancing existing computer vision algorithms to improve their performance.

As a result of research, we have implemented a demo application for analyzing movements of different kinds. Its work can be divided into 3 stages:

  • Creating the visual shell of a human by the images from multiple cameras
  • Matching the shell to a human skeleton
  • Detecting movements by a series of skeletons

Specification

Client: Intel, Nizhny Novgorod, Russia
Area of use: sports analytics
Type (platform): Research project, C++-library
Technologies and algorithms in use: C++, OpenCV, PCL, 3D reconstruction, Markerless Motion Capture, Human Action Recognition with Skeleton information

the development of algorithms for non-blocking and rarefied collective MPI-operations

the analysis of WRF, MILC and HPC Repast packages to run and work efficiently on the computing cluster Petastream consisting of computing nodes of Xeon Phi coprocessors