It takes skill to fly a drone smoothly and without glitches. And once you master the flight, you
Rogerio Bonatti, PhD candidate at the Institutein robotics from Carnegie Mellon University, one of the team members at the University of Sao Paulo, which, together with Facebook AI Research, developed a model that allows a drone to capture video based on the desired emotion or reaction of the viewer.
The drone uses angles, speeds, and flight paths to create video that can be immersive, calm, enjoyable, or unnerving, depending on what the director says.
In order to train the drone, researchersit took hundreds of videos that they tested on a thousand viewers to use to gather data on what makes the video evoke certain emotions or feelings.
The researchers then used the data to train a model that guided the drone to create a specific emotion.
The authors tested the model:they filmed a chase scene and dribbling in a soccer match. Afterwards, the authors asked viewers to give their opinions on how they felt while watching. As a result, the videos not only hit the right emotional tone, but also evoked varying degrees and completeness of feeling among respondents.
Read more
A powerful neural network has been created in China: it is 10 times more productive than analogues
Physicists have recreated the first substance that appeared after the Big Bang
Tiny hydrogen engine replaces fossil fuel counterparts