BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Understanding color and gloss perception with deep neural networks
  - Takuma Morimoto\, Fellow\, Department of Experimental Psychology\, Univ
 ersity of Oxford\, UK
DTSTART:20240919T130000Z
DTEND:20240919T140000Z
UID:TALK220978@talks.cam.ac.uk
CONTACT:Rafal Mantiuk
DESCRIPTION:In daily life\, visual properties such as color and gloss are 
 crucial for identifying and distinguishing objects around us. I will prese
 nt two research projects that investigated the mechanisms of color and glo
 ss perception using deep neural networks (DNNs). In the first project\, we
  used DNNs to explore why we have higher sensitivity to discriminate color
 s around orange than around purple. We hypothesized that this difference i
 s because our hue discrimination ability is optimized for frequently seen 
 colors in the real world. We approached this by investigating whether simi
 lar threshold patterns spontaneously emerge from DNNs trained on over a mi
 llion real-world color images. We found that human-like hue discrimination
  ability spontaneously emerges\, especially from shallower layers of the n
 etwork\, suggesting that the asymmetry in human color discrimination stems
  from chromatic regularities in the real world. The second project aimed t
 o identify a computational mechanism underpinning human gloss perception. 
 We conducted online experiments to collect large-scale human gloss judgmen
 ts for 3\,888 object images generated under diverse lighting conditions. W
 e then trained shallow neural networks using these images labeled by human
  judgments to replicate the judgment patterns. By interpreting the interna
 l architecture that emerged in the network after training\, we discovered 
 that a simple computation accounts for human gloss perception. Both projec
 ts leverage the power of machine learning to reveal novel underlying mecha
 nisms for the perception of material properties\, offering insights that t
 raditional hand-crafted models have not revealed.\n\nZoom link: https://ca
 m-ac-uk.zoom.us/j/87094914760?pwd=yBmfePQVlg50QNvA9HXvEa9e6Pc4vI.1\n
LOCATION:FW11 - William Gates Building
END:VEVENT
END:VCALENDAR