Fluendo is a leading company on multimedia solutions based on the GStreamer framework. We develop from high-end video and audio codecs, to slick end-user applications or multimedia middle-ware powering different playback, encoding and streaming scenarios. Our software is being used in millions of devices and our clients vary from the main thin client manufacturers to the world’s leading film production labels.
Job description
Fluendo has a dedicated Research Development and Innovation department working on analyzing ideas for new products and applied research in the field of multimedia applications. Our R&D projects have a direct technology transfer to final products, such as in this case with LongoMatch.
LongoMatch is a video analysis software for sports that helps coaches and sports analysts to code, analyze and present videos for games and training sessions. Our mission is to democratize video analysis in sports, providing an intuitive and powerful video analysis solution that can be adapted to any sport. LongoMatch is used worldwide in amateur and professional teams from a wide range of sports, as well as students in the field of sports analysis.
LongoMatch is a product of Fluendo, a techonology company founded in 2004 that create multimedia solutions based on the multiplatform GStreamer framework.
We are looking for a research engineer (internship) with interest in computer vision and deep learning algorithms for multimedia applications.
The goal of this project is to create an automated video tracking system for sports games. This new system will be a mix of traditional and deep learning computer vision techniques able to detect and track players and the ball in short video sequences with oclusions, as well as performing automatic camera calibration to measure distances between players and fix drawing in the game field.
The successful applicant will join our R&D team exploring novel techniques of computer vision for video segmentation, object detection and multi-target tracking.
Responsibilities
You main tasks will be:
- Support the team in develop an algorithm of automated camera calibration for football, basketball, handball and field hockey.
- Support the team in develop an algorithm for players and ball detection optimized for sports broadcast and wide-angle cameras.
- Support the team in develop an algorithm to track players effective with oclusions.
- Support the team in develop an algorithm to smooth trayectories.
- Train models to improve their results with the data sets provided by our users.
Perks
- Career path
- Beautiful offices in Barcelona's city center
- Great work environment
- Flexible working hours
- Internal hackathons
- Internal tech talks
- Free coffee
Requirements
Our ideal candadiate should have the following skills:
- Interest in computer vision and and deep learning algorithms applied to video
- Interest in deep learning frameworks such as Pytorch or Tensorflow
- Knoledge in high-level programming language like Python, C# or Java
- Fluent in English
- Proactive and autonomous
- Knowledge in multimedia frameworks like GStreamer or FFmpeg
- Ability to successfully lead and complete projects starting from scratch
