Compare human visual judgments with image representations.

<p>(A) Image-feature RDMs of 72 stimuli created from pre-trained models: LPIPS (perceptual image similarity), DINO-ViT-s8 (self-supervised model), OpenCLIP-ViT-L/14 (visual-semantic model). The elements in the RDM follow the same order as the Vision and Text RDMs. (B) Spearman’s correlation be...

وصف كامل

محفوظ في:

التفاصيل البيبلوغرافية
المؤلف الرئيسي:	Chenxi Liao (5622617) (author)
مؤلفون آخرون:	Masataka Sawayama (5150747) (author), Bei Xiao (14432449) (author)
منشور في:	2024
الموضوعات:	Neuroscience Science Policy Biological Sciences not elsewhere classified two behavioral tasks discover structural differences visual judgments exhibit visual features relate communicate relevant information unsupervised alignment method individual differences compared form verbal descriptions verbal descriptions visual nuances visual judgment leveraging information xlink "> work illustrates wide range visually discriminate various pre similarity structures significant correlation results show novel framework material perception material appearances known categories human cognition human behaviors generate images fully convey findings reveal different modalities comprehensive model coarse level categorical level based models ambiguous categories
الوسوم:	إضافة وسم لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!

Compare human visual judgments with image representations.

مواد مشابهة