Deep Learning to Help Preserve Privacy from Wearable Cameras

Wearable cameras are opening up exciting new applications, but will also require new techniques to help people preserve their privacy.

A group of researchers from Indiana University and Olin College used the Caffe deep learning framework and a Tesla K20 GPU to automatically detect private content on monitors that people may not want to be recorded.

Enhancing lifelong privacy by detecting screens
Random images from an author’s lifelog, showing that computer and phone displays with private information are common.

As a first step, they trained their model to detect computer monitors and phone screens, which achieved 99.8% near-human accuracy. They then trained a Convolutional Neural Network to classify whether each screen is displaying a privacy-sensitive application such as Facebook, GMail, Mac OS Messages, or other similar applications – and their solution achieved 54.2% accuracy even though many of the screens were not fully visible.

Read the research paper >>

About Brad Nemire

Brad Nemire
Brad Nemire is on the Developer Marketing team and loves reading about all of the fascinating research being done by developers using NVIDIA GPUs. Reach out to Brad on Twitter @BradNemire and let him know how you’re using GPUs to accelerate your research. Brad graduated from San Diego State University and currently resides in San Jose, CA. Follow @BradNemire on Twitter