CVPR2017読み会

今回は、2017年7月に行われたConference on Computer Vision and Pattern Recognition 2017(CVPR2017)にて発表された研究2本を紹介します。

Transcript

1. $713ಡΈձ
   %XBOHP.FEJB7JMMBHF
   IJSPTIJCBL
   
   

   View Slide

2. ঺հ͢Δ࿦จ
   %FFQ-FBSOJOHܥ
   
   4UBDLFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLT
   ڭࢣ͋ΓֶशλεΫͷதؒදݱΛੜ੒͢Δֶशͱɺ
   தؒදݱ͔Βը૾Λੜ੒͢ΔֶशΛಉ࣌ʹߦ͏ֶशख๏
   )JHI3FTPMVUJPO*NBHF*OQBJOUJOHVTJOH.VMUJ4DBMF/FVSBM1BUDI4ZOUIFTJT
   ࠷ॳʹ௿ղ૾౓ͳը૾Ͱम෮ͯ͠ߴղ૾౓ʹ֦େ͠ɺ
   पғͷը૾ͱ7((ಛ௃ྔ͕ۙ͘ͳΔΑ͏ʹೖྗը૾Λ࠷దԽ͢Δख๏
   'JHD
   'JH
   

   View Slide

3. 4UBDLFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLT
   
   Xun Huang1 Yixuan Li2 Omid Poursaeed2 John Hopcroft1 Serge Belongie1,3
   1Department of Computer Science, Cornell University
   2School of Electrical and Computer Engineering, Cornell University
   3Cornell Tech
   BEWFSTBSJBMMPTT
   &ODPEFS
   ʢ෼ྨλεΫΛࣄલֶशʣ
   (FOFSBUPS
   %JTDSJNJOBUPS
   2OFUXPSL QSFEJDU[
   
   ʢ΄ͲΜͲ%ͱಉ͡ߏ଄ʣ
   DPOEJUJPOBMMPTT
   ʢ࠶ߏ੒ʣ
   FOUSPQZMPTT
   ʢϊΠζ[Λແࢹ͠ͳ͍Α͏ʹʣ
   ߴ࣍ಛ௃ྔΛ࢖ͬͯΑΓ௿࣍ͳಛ௃ྔΛ࡞Δ
   'JH
   

   View Slide

4. 4UBDLFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLT
   
   'JH
   'JH
   

   View Slide

5. 4UBDLFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLT
   
   

   View Slide

6. )JHI3FTPMVUJPO*NBHF*OQBJOUJOHVTJOH.VMUJ4DBMF/FVSBM1BUDI
   4ZOUIFTJT
   
   Chao Yang¡1, Xin Lu§2 , Zhe Lin†2 , Eli Shechtman‡2 , Oliver Wang∗2 , and Hao Lik1,3,4
   1University of Southern California
   2Adobe Research
   3Pinscreen
   4USC Institute for Creative Technologies
   ௿ղ૾౓ͳը૾Λ%FFQͰੜ੒͠ɺ
   म෮ՕॴͷपΓͷը૾ʹ߹͏Α͏ʹ/FVSBM1BUDI
   'JH
   

   View Slide

7. )JHI3FTPMVUJPO*NBHF*OQBJOUJOHVTJOH.VMUJ4DBMF/FVSBM1BUDI
   4ZOUIFTJT
   
   'JH
   'JH
   

   View Slide