Fingerprints of Super Resolution Networks
This work addresses the problem of model identification and security for SISR networks, which is incremental as it extends prior research on GAN fingerprints to a more challenging class of image generation models.
The paper investigates whether single image super-resolution (SISR) networks leave identifiable fingerprints in their output images, similar to GANs, and finds that networks with high upscaling factors or trained with adversarial loss produce distinctive fingerprints, enabling reverse-engineering of some hyperparameters under certain conditions.
Several recent studies have demonstrated that deep-learning based image generation models, such as GANs, can be uniquely identified, and possibly even reverse-engineered, by the fingerprints they leave on their output images. We extend this research to single image super-resolution (SISR) networks. Compared to previously studied models, SISR networks are a uniquely challenging class of image generation model from which to extract and analyze fingerprints, as they can often generate images that closely match the corresponding ground truth and thus likely leave little flexibility to embed signatures. We take SISR models as examples to investigate if the findings from the previous work on fingerprints of GAN-based networks are valid for general image generation models. We show that SISR networks with a high upscaling factor or trained using adversarial loss leave highly distinctive fingerprints, and that under certain conditions, some SISR network hyperparameters can be reverse-engineered from these fingerprints.