As technology becomes more sophisticated and internet connections become ubiquitous the need to be physically present for meetings quickly becomes a thing of the past. This happens not only in business, but also in education. Being able to follow lectures remotely greatly improves the way students can enjoy their education. Not only do lecture recordings make it possible to view a lecture from anywhere, it can also lead to a system where students can follow any lecture given in any part of the world.
For SURFnet we investigated the possibilities of using computer vision to make low cost unattended lecture capturing solutions.
As part of our investigation into the possibilities and available technologies, we developed a rough prototype of a working solution for lecture capture. This prototype consists of 4 tools:
- A calibration tool, which analyses video from an empty lecture room to eliminate errors in the detection of the presenter.
- A detection tool, which detects and tracks presenters.
- A direction tool, which acts as an automated film director, translating the position and movement of the speaker into smooth camera directions.
- A processing tool, which takes the instructions from the director and controls a physical camera (for live recordings) or edits video (for post-processing existing recordings).
More details about the project can be found here: