"Privacy-Preserving Camera Captures Only the Objects You Want"

Digital cameras are present in embedded in smartphones, smart eyewear, security surveillance systems, autonomous vehicles, facial recognition technologies, and more. As the volume of image data captured by cameras has increased, so have concerns about privacy protection. Researchers have explored the possibility of taking photos that instantly capture only the objects of interest in a frame while obscuring unnecessary or potentially sensitive details, without the need for any editing, encryption, or other digital post-processing. A team of researchers at the University of California, Los Angeles (UCLA) recently released a paper describing a smart camera informed by Artificial Intelligence (AI) that accomplishes that. Their device is made up of several transmissive, 3D-printed surfaces, each of which has tens of thousands of diffractive features at the scale of a light wavelength. The team designed and assembled the successive layers using deep learning-based training. Although the technology is still in the early stages, its ability to erase sensitive information at the point of capture could eliminate the risk of hacking or the exposure of original, unedited versions of images, particularly those stored in the cloud. Simultaneously, it eliminates the need for any post-processing, which is data-intensive and uses more power. This article continues to discuss the privacy-preserving camera designed by researchers from the UCLA Samueli School of Engineering. 

UCLA reports "Privacy-Preserving Camera Captures Only the Objects You Want"