ImageNet-trained CNNs are biased towards texture ← Katedra kybernetiky

Nikos Efthymiadis presents ImageNet-trained CNNs are biased towards texture

On 2019-11-21 - 2019-11-21 11:00:00 at G205, Karlovo náměstí 13, Praha 2
Reading group on the work "ImageNet-trained CNNs are biased towards texture;
increasing shape bias improves accuracy and robustness" (ICLR 2019) by
R. Geirhos, P. Rubisch, C. Michaelis, M. Bethge, F. Wichmann, and W. Brendel,
presented by Nikos Efthymiadis.

Paper abstract: Convolutional Neural Networks (CNNs) are commonly thought to
recognise objects by learning increasingly complex representations of object
shapes. Some recent studies suggest a more important role of image textures. We
here put these conflicting hypotheses to a quantitative test by evaluating CNNs
and human observers on images with a texture-shape cue conflict. We show that
ImageNet-trained CNNs are strongly biased towards recognising textures rather
than shapes, which is in stark contrast to human behavioural evidence and
reveals fundamentally different classification strategies. We then demonstrate
that the same standard architecture (ResNet-50) that learns a texture-based
representation on ImageNet is able to learn a shape-based representation
instead when trained on "Stylized-ImageNet", a stylized version of ImageNet.
This provides a much better fit for human behavioural performance in our
well-controlled psychophysical lab setting (nine experiments totalling 48,560
psychophysical trials across 97 observers) and comes with a number of
unexpected emergent benefits such as improved object detection performance and
previously unseen robustness towards a wide range of image distortions,
highlighting advantages of a shape-based representation.

Paper URL: https://arxiv.org/abs/1811.12231

Instructions for participants: The reading group studies the literature in the
field of pattern recognition and computer vision. At each meeting one or more
papers are prepared for presentation by a single person, the presenter. The
meetings are open to anyone, disregarding their background. It is assumed that
everyone attending the reading group has, at least briefly, read the paper –
not necessarily understanding everything. Attendants should preferably send
questions about the unclear parts to the speaker at least one day in advance.
During the presentation we aim to have a fruitful discussion, a critical
analysis of the paper, as well as brainstorming for creative extensions.

See the page of reading groups
http://cmp.felk.cvut.cz/~toliageo/rg/index.html

The reading group was moved from 19th Nov. to 21st Nov. due to overlap with
https://dcgi.fel.cvut.cz/en/events/main#invited-talk-ariel-shamir-the-face-of-artlandmark-detection-geometric-style-in-portraits
Za obsah zodpovídá: Petr Pošík