Detect Fake Physical Tags With Deepfake

September 02, 2022
Detect Fake Physical Tags With Deepfake

Deepfake is an AI technology that digitally alters a person’s face so that they appear to be someone else. It is often used with bad intentions. We turned the tables and found a use for good. Blocktag explored using deepfake to digitally alter an original tag so that you can anticipate how photocopied or reprinted tags look like.

Deep Learning For Print Physics

Blocktag’s deepfake technology learns the physics of printing to predict how reprinted tags will look like. Below is an image rendering of deepfake reprints.
DeepfakeTagSample1
Figure 1: Deepfake reprints derived from an original tag.

Blocktag can also use AI to customize how deepfake tags look like on different phone cameras. Because cameras with different resolutions perceive the printed original tag at different sharpness.

DeepfakeTagSample2
Figure 2: Deepfake reprints derived from the same original tag taken with a higher resolution camera.

Blocktag uses deepfakes to improve authentication accuracy. As we collect more training data from user authentications, there are more deepfake datapoints to train deep learning models with higher accuracies.

Deepfake empowers consumers as anti-counterfeiting's first line of defense. Before counterfeiters try to fake it till they make it.

ChunMing Chin
Cofounder, Director of Engineering
Blocktag Inc