English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Humans rely heavily on shapes as a primary cue for object recognition. Assecondary cues, colours and textures are also beneficial in this regard.Convolutional neural networks (CNNs), an imitation of biological neuralnetworks, have been shown to exhibit conflicting properties. Some studiesindicate that CNNs are biased towards textures whereas, another set of studiessuggests shape bias for a classification task. However, they do not discuss therole of colours, implying its possible humble role in the task of objectrecognition. In this paper, we empirically investigate the importance ofcolours in object recognition for CNNs. We are able to demonstrate that CNNsoften rely heavily on colour information while making a prediction. Our resultsshow that the degree of dependency on colours tend to vary from one dataset toanother. Moreover, networks tend to rely more on colours if trained fromscratch. Pre-training can allow the model to be less colour dependent. Tofacilitate these findings, we follow the framework often deployed inunderstanding role of colours in object recognition for humans. We evaluate amodel trained with congruent images (images in original colours eg. redstrawberries) on congruent, greyscale, and incongruent images (images inunnatural colours eg. blue strawberries). We measure and analyse network'spredictive performance (top-1 accuracy) under these different stylisations. Weutilise standard datasets of supervised image classification and fine-grainedimage classification in our experiments.

Assessing The Importance Of Colours For CNNs In Object Recognition