Image-based Gimbal Control

The Shift extension is a practical example of an image-based motion control system for drones. The system is mainly designed to track distanced objects (whole body / object shots). However for different cinematographic shots (medium shot, close up, etc) different tracking methods are needed [1].

This thesis focusses on the cinematographic needs and aims at tracking based on shot types. The to be developed system should extend the tracking abilities of state of the art image-based control systems in terms of shot types variety. A live video stream of a DSLR mounted on a camera gimbal (a stabilizing system such as the Ronin) is going to be analyzed and based on this analysis the physical camera motion is controlled.

[1] D. Hulens, T. Goedemé, and T. Rumes, "Autonomous lecture recording with a PTZ camera while complying with cinematographic rules" in Computer and Robot Vision (CRV), 2014 Canadian Conference on, 2014, pp. 371-377.

• Literature review on related work
• Computational analysis of live video feed (applying frameworks like OpenCV)
• Remote control of camera gimbal motion (via bluetooth)
• Written thesis and presentation of your work