Training with the Invisibles: Obfuscating Images to Share Safely for Learning Visual Recognition Models
This addresses privacy concerns in data sharing for computer vision, enabling safer collaboration without significant performance loss, though it is incremental as it builds on existing obfuscation and recognition methods.
The paper tackles the problem of sharing visual data for training recognition models while preserving privacy by obfuscating images to be unrecognizable to humans but usable by machines. It shows that models trained on obfuscated data perform within about 1% margin (up to 0.48%) of those trained on original data across tasks like image classification and facial landmark detection.
High-performance visual recognition systems generally require a large collection of labeled images to train. The expensive data curation can be an obstacle for improving recognition performance. Sharing more data allows training for better models. But personal and private information in the data prevent such sharing. To promote sharing visual data for learning a recognition model, we propose to obfuscate the images so that humans are not able to recognize their detailed contents, while machines can still utilize them to train new models. We validate our approach by comprehensive experiments on three challenging visual recognition tasks; image classification, attribute classification, and facial landmark detection on several datasets including SVHN, CIFAR10, Pascal VOC 2012, CelebA, and MTFL. Our method successfully obfuscates the images from humans recognition, but a machine model trained with them performs within about 1% margin (up to 0.48%) of the performance of a model trained with the original, non-obfuscated data.