ȸ¿øµî·Ï £ü ºñ¹øºÐ½Ç
ȯ°æ,°æÁ¦,»ýÅÂ



4Â÷¿ø ½º¸¶Æ®½ÃƼ



ÇÁ·ÓÅ×Å©  

¤ýÀÛ¼ºÀÚ ½º¸¶Æ®¿ÀÅä½ÃƼ
¤ýÀÛ¼ºÀÏ 2016³â 11¿ù 26ÀÏ Åä¿äÀÏ
¤ýȨÆäÀÌÁö http://cpagent.com
¤ýLink#1 google-deepmind-ai-lip-reading-tv (Down:132)
¤ýÃßõ: 0  ¤ýÁ¶È¸: 3074      
±¸±ÛÀÇ ÀΰøÁö´ÉÆÀÀÎ 'µö¸¶Àεå(DeepMind)'¿Í '¿Á½ºÆ÷µå´ëÇÐ(Oxford)' ¿¬±¸ÁøÀÌ »ç¶÷ÀÇ ÀÔ¸ð¾çÀ» ºÐ¼®Çس»´Â ¼ÒÇÁÆ®¿þ¾î¸¦ °³¹ß

±¸±ÛÀÇ ÀΰøÁö´ÉÆÀÀÎ 'µö¸¶Àεå(DeepMind)'¿Í '¿Á½ºÆ÷µå´ëÇÐ(Oxford)' ¿¬±¸ÁøÀÌ »ç¶÷ÀÇ ÀÔ¸ð¾çÀ» ºÐ¼®Çس»´Â ¼ÒÇÁÆ®¿þ¾î¸¦ °³¹ßÇß´Ù°í ÇÕ´Ï´Ù




http://www.theverge.com/¡¦/13740798/google-deepmind-ai-lip-r¡¦





Google¡¯s AI can now lip read better than humans after watching thousands of hours of TV

The AI system annotated TV footage with 46.8 percent accuracy

Researchers from Google¡¯s AI division DeepMind and the University of Oxford have used artificial intelligence to create the most accurate lip-reading software ever. Using thousands of hours of TV footage from the BBC, scientists trained a neural network to annotate video footage with 46.8 percent accuracy. That might not seem that impressive at first — especially compared to AI accuracy rates when transcribing audio — but tested on the same footage, a professional human lip-reader was only able to get the right word 12.4 percent of the time.




The research follows similar work published by a separate group at the University of Oxford earlier this month. Using related techniques, these scientist were able to create a lip-reading program called LipNet that achieved 93.4 percent accuracy in tests, compared to 52.3 percent human accuracy. However, LipNet was only tested on specially-recorded footage that used volunteers speaking formulaic sentences. By comparison, DeepMind¡¯s software — known as ¡°Watch, Listen, Attend, and Spell¡± — was tested on far more challenging footage; transcribing natural, unscripted conversations from BBC politics shows.




More than 5,000 hours of footage from TV shows including NewsnightQuestion Time, and the World Today, was used to train DeepMind¡¯s ¡°Watch, Listen, Attend, and Spell¡± program. The videos included 118,000 difference sentences and some 17,500 unique words, compared to LipNet¡¯s test database of video of just 51 unique words.




DeepMind¡¯s researchers suggest that the program could have a host of applications, including helping hearing-impaired people understand conversations. It could also be used to annotate silent films, or allow you to control digital assistants like Siri or Alexa by just mouthing words to a camera (handy if you¡¯re using the program in public).




But when most people learn that an AI program has learned how to lip-read, their first thought is how it might be used for surveillance. Researchers say that there¡¯s still a big difference in transcribing brightly-lit, high resolution TV footage, and grainy CCTV video with a low frame rate, but you can¡¯t ignore the fact, that artificial intelligence seems to be closing this gap.




   
  0
3500
   
¹øÈ£     ±Û Á¦ ¸ñ  ÀÛ¼ºÀÚ ÀÛ¼ºÀÏ Á¶È¸
311 ½º¸¶Æ®½ÃƼ¶ó´Â Ç÷§Æû) Á¤ºÎ, ½º¸¶Æ®½ÃƼ ½Ã¹üµµ½Ã '¼¼Á¾¡¤ºÎ»ê.. ½º¸¶Æ®¿ÀÅä½ÃƼ 2018-01-30 3774
310 »ï¼º, Ç® ºê¶ó¿ì¡ ±â´É °®Ãá '¸ÖƼ ÅÍÄ¡Æù' Ãâ½Ã 3Â÷¿ø¼îÇθô 2008-01-25 3742
309 SKT¡¤¹Ì±¹ ¸³¸ð¼Ç, ½Ç°¨Çü ¸ÖƼ¹Ìµð¾î ¼­ºñ½º °øµ¿°³¹ß MOU ü°á.. ½º¸¶Æ®µµ½Ã 2015-10-17 3729
308 [½º¸¶Æ®½ÃƼ] U-City °Ç¼³¹ý ¡æ ¡¸½º¸¶Æ®µµ½Ã Á¶¼º ¹× »ê¾÷ÁøÈï¹ý.. ½º¸¶Æ®¿ÀÅä½ÃƼ 2017-03-03 3705
307 ÈÞ´ëÆù Ç®ºê¶óÀÌ¡ ³×À̹ö À¥¼­ÇÎ Á¤´äÀº °¡º­¿î Ŭ¸¯°¡´É »óÈ£ÀÛ.. [1] 3dCTÆÀ 2008-01-25 3701
306 ³ì»ö¼ºÀå Á¤Ã¥ÀÇ º£½ºÆ® ÇÁ·¢Æ¼½º ¿¡ÄÚÀ̸ÞÁö³×ÀÌ¼Ç 2009-08-07 3661
305 [Áö¹æ¼Ò¸ê º¸°í¼­] ¨ê "¿ì¸° ÀÌ·¸°Ô ±Øº¹Çß¾î¿ä"..ÀÏÀÚ¸®°¡ ºñ°á.. ½º¸¶Æ®µµ½ÃÀç»ý 2017-07-12 3632
304 °Ç¼³»ç ÁÙµµ»ê °øÆ÷ `ÀÏÆĸ¸ÆÄ` [5] 3dCTÆÀ 2008-07-15 3494
303 °í¾ç½Ã, ¼öµµ±Ç ¼­ºÏºÎ ¡®¸£³×»ó½º¡¯ °³¸· 4Â÷¿øÀÎÅͳݵµ½Ã 2008-09-04 3485
302 [ÈÞ´ëÆùÇ®ºê¶ó¿ì¡]ÈÞ´ëÆùÀ¸·Î º¸´Â ¼¼»ó, 4¹è ³Ð¾îÁø´Ù [1] 3Â÷¿ø¼îÇθô 2008-01-25 3482
301 ºÎ»ê¼­ ICT Æ®·»µå °øÀ¯ÀÇ Àå ¿­¸°´Ù 'U-IoT ¿ùµå ÄÁº¥¼Ç' cpagent 2015-12-01 3444
300 ž籤¡¤Ç³·Â¡¤LED¡¤Àü·ÂIT¡¦¡®³ì»ö µµ·Î¡¯ ´Þ¸°´Ù ¾Å¾Å~ ¿¡ÄÚÀ̸ÞÁö³×ÀÌ¼Ç 2008-10-06 3393
299 ÅÍÄ¡, ÇÝƽ µî ÃֽŠ¸ð¹ÙÀÏ UX °øÀ¯ÀÇ Àå ¿­·Á 4Â÷¿øÀÎÅͳݵµ½Ã 2008-04-22 3373
298 3¿ù 11ÀÏ ÀüÀڽŹ®] "¸ðµ¨ÇϿ콺µµ °¡»óÇö½Ç·Î ´ëüÇÑ´Ù" 3Â÷¿øÀÎÅͳݵµ½Ã 2008-03-31 3342
297 ºÐ¾ç°¡°¡ 'ÃÖ°í °æÀï·Â'ÀÌ´Ù 3D¸ðµ¨ÇϿ콺 2008-08-09 3302
296 SKT¡¤BMW, ¼¼°è ÃÖÃÊ 5G Ä¿³ØƼµåÄ« ½Ã¿¬ ½º¸¶Æ®¿ÀÅä½ÃƼ 2016-11-15 3237
295 080311] "¸ðµ¨ÇϿ콺µµ °¡»óÇö½Ç·Î ´ëüÇÑ´Ù" [3] 3dCTÆÀ 2008-06-27 3223
294 [ÈÞ´ëÆù Ç®ºê¶óÀÌ¡] ÈÞ´ëÆùÀ¸·Î À¯¼± À¥ ¼­ºñ½º ¶È°°ÀÌ ÀÌ¿ëÇÑ´Ù.. [1] 3Â÷¿ø¼îÇθô 2008-01-25 3130
293 ±¸±ÛÀÇ ÀΰøÁö´ÉÆÀÀÎ 'µö¸¶Àεå(DeepMind)'¿Í '¿Á½ºÆ÷µå´ëÇÐ(Oxf.. ½º¸¶Æ®¿ÀÅä½ÃƼ 2016-11-26 3074
292 ¹Ì·¡ºÎ¡¤NIPA, 5ÀϺÎÅÍ °³¹æÇü ½º¸¶Æ®½ÃƼ °úÁ¦ °ø¸ð ½º¸¶Æ®¿ÀÅä½ÃƼ 2016-04-04 3069
12345678910,,,20