[Preview] Demonstrating InfraredTags: Decoding Invisible 3D Printed Tags with Convolutional Neural Networks
Mustafa Doga Dogan, Veerapatr Yotamornsunthorn, Ahmad Taka, Yunyi Zhu, Aakar Gupta, Stefanie Mueller
CHI'22: ACM Conference on Human Factors in Computing Systems
Session: Interactivity
Abstract
We demonstrate InfraredTags, which are invisible 2D markers and barcodes that are 3D printed as part of objects. We show how InfraredTags can be decoded using a convolutional neural network (CNN) after being captured by low-cost near-infrared cameras. InfraredTags are formed by printing objects from an infrared-transmitting filament, which infrared cameras can see through, and by having air gaps inside for the tag's bits, which appear at a different intensity in the infrared image. We built a user interface that facilitates the integration of common tags (QR codes, ArUco markers) with the object geometry to make them 3D printable as InfraredTags.
Once printed, the tags can be robustly decoded with a U-Net model that we trained using a custom dataset for optimal binarization and detection. We show how our method enables different applications, such as object tracking and embedding metadata for tangible interactions and augmented reality.
WEB:: https://programs.sigchi.org/chi/2022/program/content/72083
Presentation Video:: https://www.youtube.com/watch?v=TcPwWlTlDYc
DOI:: https://doi.org/10.1145/3491101.3519905
Video previews for CHI 2022 Interactivity