Blind Vision
We have developed a general framework for secure image and video analysis that allows a client to have his data analyzed by a server, privately. For example, the client might submit his images to the server for face detection, without letting the server learn anything about the content of the images. Or, more generally, the client might use a query image to query an image database stored on the server, without revealing the content of the query image to the server. In the last year, we have implemented a secure face detector as a proof-of-concept, presented our work at a scientific conference and extended the method to work with different types of machine learning technologies. We have extended the blind vision framework in two directions: 1) We have explored ways to accelerate the process, using domain specific knowledge; and 2) We have extended blind vision to handle general image matching problems.
Background & Objective: The problem of image and video analysis has been extensively investigated within the computer vision community. However, privacy concerns were never taken into account, as the assumption was always that the data is available to the analyzing algorithm. We borrow techniques from the field of Secure Multi-Party Computations to derive secure image analysis algorithms.
Technical Discussion: Our work combines methods from Computer Vision, Machine Learning and Secure Multi Party Computations. We use cryptographic primitives such as Oblivious Transfer to convert non-secure operations into a secure one. To this end, we have shown (i) how to accelerate a secure face detection algorithms and (ii) how to perform general image matching algorithms in a secure fashion.
Future Direction: We plan to keep exploring other computer vision and machine learning applications that will benefit from a secure framework.
Contact: Joseph Katz
| Technical Reports: | |
| Blind Vision | |
Technology Areas:
Imaging
Computer Vision
Modification Date: August 31, 2007
