Computer Vision Art Gallery

CVPR 2021: Virtual due to COVID-19

Bernat CunĂ­

C is for CAT, (2020)

Project Website

C4CAT is a visual dissection of popular datasets used by ML tools, made to expose genericness and claim for hyper-localized datasets.

Technically: it a set of compositions of the 80 categories from the COCO dataset, made by running the Mask R-CNN for Object Detection and Segmentation model through the dataset, and compose the identified cut-outs into object-mosaics to nicely visualize the overall essence of each category.

Back To The Gallery