Scientists created about 80 hours of video with 8 000 000 frames. They contain 1 109 manually labeled microexpressions and 3 490
As important non-verbal cuesmicroexpressions communications can reveal a person's true emotional state. Their analysis is now attracting the attention of experts due to the development of deep learning methods. However, research in both psychology and artificial intelligence has been limited to a small sample of microexpressions.
Data size is vital forthe use of artificial intelligence to conduct microexpression analysis, and the large sample size allows for effective testing of the microexpression analysis method, avoiding database bias, the scientists note.
One of the advantages of the new base is its highcredibility. During the identification process, the researchers optimized the data collection environment to be as close to reality as possible, and participants participated in a mock crime and interrogation. As part of this process, video, voice, and physiological signals were recorded using a video recorder and wearable devices , including electrodermal activity, heart rate/fingertip pulse, respiration, and pulse.
The experiment showed that subjectsIn “high stakes” scenarios (e.g., committing a crime), more microexpressions are missed than in “low stakes” scenarios (e.g., innocence). Thus, high-fidelity samples provide the basis for robust real-world microexpressions analysis and emotion understanding.
Read more
American satellite "saw" an unusual message from Earth
Published video from the rocket, which was launched from an experimental accelerator
The monster at the center of our Galaxy: look at the photo of a black hole in the Milky Way