科学美国人60秒：计算机能分辨真笑还是假笑
燕山大学 刘立军 供稿
【TRANSCRIPT】
This is Scientific American's 60-second Science, I'm Sophie Bushwick.
Is that person really glad to see me? Or are they just being polite? Some people struggle to distinguish a perfunctory grin from a truly happy smile. And computers have found this task even more difficult - that is, until researchers trained a program to detect when a smile is genuine.
Visual computing researchers at the University of Bradford in the U.K. started with software for analyzing a changing facial expression. This program can examine a video clip of a human head and identify specific details around the eyes, cheeks and mouth. Then the program tracks the details moving relative to each other as the face smiles.
Next, the scientists had their program evaluate two sets of video clips. In one, subjects performed posed smiles. In the other, they watched a film that inspired genuine displays of emotion. The program calculated the differences among the subjects' faces during the two clips. And it turns out that your mouth, cheeks and eyes move differently when you're faking that smirk.
In particular, the muscles around the eyes shift 10 percent more for a real smile than they do for a fake one. These results are in the journal Advanced Engineering Informatics.
The researchers suggest their work could improve a computer's ability to analyze facial expressions and thus to interact more smoothly with humans. But their real accomplishment is in proving Tyra Banks right: "You have to smile with your eyes."
Thanks for listening for Scientific American - 60-Second Science. I'm Sophie Bushwick.
Adapted from http://www.kekenet.com/broadcast/202006/611612.shtml
【VOCABULARY】
1. perfunctory adj. (formal) (of an action 行为) done as a duty or habit, without real interest,
attention or feeling 敷衍的；例行公事般的；潦草的。例如：a perfunctory nod / smile敷衍的点头 / 微笑
2. smirk v. to smile in a silly or unpleasant way that shows that you are pleased with yourself, know sth. that other people do not know, etc. 自鸣得意地笑；傻笑。例如：It was hard not to smirk.
让人忍俊不禁。
【QUESTIONS】
Read the passage. Then listen to the news and fill in the blanks with the information (words, phrases or sentences) you hear.
This is Scientific American's 60-second Science, I'm Sophie Bushwick.
Is that person really glad to see me? Or are they just being polite? Some people struggle to (Q1) __________________________________________. And computers have found this task even more difficult - that is, until researchers trained a program to (Q2) _________________ when a smile is genuine.
Visual computing researchers at the University of Bradford in the U.K. started with (Q3) _________ for analyzing a changing (Q4) __________________________. This program can examine a video clip of a human head and identify specific details around the eyes, cheeks and mouth. Then the program tracks the details moving relative to each other as the face smiles.
Next, the scientists had their program evaluate (Q5) _____________________. In one, subjects performed posed smiles. In the other, they watched a film that inspired genuine displays of emotion. The program (Q6) _____________________ the differences among the subjects' faces during the two clips. And it turns out that your mouth, cheeks and eyes move (Q7) _____________________ when you're faking that smirk.
In particular, the muscles around the eyes shift (Q8) ___________________________ for a real smile than they do for a fake one. These results are in the journal Advanced Engineering Informatics.
The researchers suggest their work could (Q9) ___________________ a computer's ability to analyze facial expressions and thus to interact more smoothly with humans. But their real accomplishment is in proving Tyra Banks right: "You have to (Q10) ___________________."
Thanks for listening for Scientific American - 60-Second Science. I'm Sophie Bushwick.
【KEY】
Read the passage. Then listen to the news and fill in the blanks with the information (words, phrases or sentences) you hear.
This is Scientific American's 60-second Science, I'm Sophie Bushwick.
Is that person really glad to see me? Or are they just being polite? Some people struggle to (Q1) distinguish a perfunctory grin from a truly happy smile. And computers have found this task even more difficult - that is, until researchers trained a program to (Q2) detect when a smile is genuine.
Visual computing researchers at the University of Bradford in the U.K. started with (Q3) software for analyzing a changing (Q4) facial expression. This program can examine a video clip of a human head and identify specific details around the eyes, cheeks and mouth. Then the program tracks the details moving relative to each other as the face smiles.
Next, the scientists had their program evaluate (Q5) two sets of video clips. In one, subjects performed posed smiles. In the other, they watched a film that inspired genuine displays of emotion. The program (Q6) calculated the differences among the subjects' faces during the two clips. And it turns out that your mouth, cheeks and eyes move (Q7) differently when you're faking that smirk.
In particular, the muscles around the eyes shift (Q8) 10 percent more for a real smile than they do for a fake one. These results are in the journal Advanced Engineering Informatics.
The researchers suggest their work could (Q9) improve a computer's ability to analyze facial expressions and thus to interact more smoothly with humans. But their real accomplishment is in proving Tyra Banks right: "You have to (Q10) smile with your eyes."
Thanks for listening for Scientific American - 60-Second Science. I'm Sophie Bushwick.