Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
½ÉÃþ ½Å°æ¸ÁÀÇ ÃÖÀûȸ¦ ÅëÇÑ ¼Ò±Ô¸ð Çൿ ºÐ·ù ¹®Á¦ÀÇ Çൿ ÀÎ½Ä ¹æ¹ý |
¿µ¹®Á¦¸ñ(English Title) |
A Method of Activity Recognition in Small-Scale Activity Classification Problems via Optimization of Deep Neural Networks |
ÀúÀÚ(Author) |
±è½ÂÇö
±è¿¬È£
±èµµ¿¬
Seunghyun Kim
Yeon-Ho Kim
Do-Yeon Kim
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 06 NO. 03 PP. 0155 ~ 0160 (2017. 03) |
Çѱ۳»¿ë (Korean Abstract) |
ÃÖ±Ù ÄÄÇ»Å͸¦ ÀÌ¿ëÇÑ ´Ù¾çÇÑ ÀÎ½Ä ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ µö ·¯´×À» Àû¿ëÇÏ´Â »ç·Ê°¡ ´Ã¾î³ª°í ÀÖ´Ù. µö ·¯´×Àº ÇнÀ¿¡ ÇÊ¿äÇÑ ¿ä¼Ò¸¦ ÇнÀµ¥ÀÌÅ͸¦ ÅëÇØ ½º½º·Î µµÃâÇس»±â ¶§¹®¿¡, ¼öÀÛ¾÷(hand-craft)À» ÅëÇØ Æ¯Â¡À» µµÃâÇÏ´ø ±âÁ¸ÀÇ ±â°èÇнÀ ¹æ¹ýº¸´Ù ´õ ¸¹Àº ÀåÁ¡À» °®´Â´Ù. ÇൿÀνÄÀ» À§ÇÑ ±âÁ¸ÀÇ ½ÉÃþ ½Å°æ¸ÁÀº ºñµð¿À µ¥ÀÌÅ͸¦ ÀÏÁ¤ ÇÁ·¹ÀÓÀÇ À̹ÌÁö·Î ºÐÇÒÇÑ ÈÄ, ºÐÇÒµÈ °¢ À̹ÌÁö »çÀÌÀÇ ½Ã°£Àû ¿¬°è¼º ºÐ¼®À» ÅëÇØ ÇൿÀ» ºÐ·ùÇÑ´Ù. ±×·¯³ª ÀÌ·¯ÇÑ ½Å°æ¸ÁÀº ¼Ò±Ô¸ð Çൿ Ŭ·¡½º¸¦ °®´Â ºÐ·ù ¹®Á¦¿¡¼ ÇнÀ µ¥ÀÌÅÍÀÇ ºÎÁ· ¹®Á¦ ¹× °úÀûÇÕ(overfitting) ¹®Á¦·Î ÀÎÇØ À̸¦ ½ÇÁ¦ ¹®Á¦¿¡ Àû¿ëÇϱ⠾î·Á¿î °æ¿ì°¡ ¸¹´Ù. ÀÌ¿¡ º» ³í¹®¿¡¼´Â 5°¡ÁöÀÇ ¼Ò±Ô¸ð Çൿ Ŭ·¡½º¸¦ Á¤ÀÇÇÏ°í, ±âÁ¸ Çൿ ÀÎ½Ä ½Å°æ¸ÁÀÇ ÃÖÀûȸ¦ ÅëÇØ À̸¦ ºÐ·ùÇÏ¿´´Ù. 700°³ÀÇ ºñµð¿Àµ¥ÀÌÅ͸¦ ÅëÇØ Çൿ µ¥ÀÌÅͺ£À̽º¸¦ ±¸¼ºÇÏ¿´°í, ¾à 74.00%ÀÇ ºÐ·ù Á¤È®µµ¸¦ ¾òÀ» ¼ö ÀÖ¾ú´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Recently, Deep learning has been used successfully to solve many recognition problems. It has many advantages over existing machine learning methods that extract feature points through hand-crafting. Deep neural networks for human activity recognition split video data into frame images, and then classify activities by analysing the connectivity of frame images according to the time. But it is difficult to apply to actual problems which has small-scale activity classes. Because this situations has a problem of overfitting and insufficient training data. In this paper, we defined 5 type of small-scale human activities, and classified them. We construct video database using 700 video clips, and obtained a classifying accuracy of 74.00%.
|
Å°¿öµå(Keyword) |
Çൿ ÀνÄ
½ÉÃþ ½Å°æ¸Á
LRCN
ÃÖÀûÈ
Activity Recognition
Deep Neural Network
LRCN
Optimization
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|