Scientists at the Inception Institute of Artificial Intelligence in the United Arab Emirates, the Beijing Institute of Technology and Stony Brook University have developed an Artificial Intelligence (AI) system that removes blur from images after they are taken. Amazingly they are able to remove blur from human faces.
The model learns to sort out what is the foreground and background and what is blurred. The team compiled a dataset called Human-aware Image Deblurring (HIDE) which has pairs of blurry images and ground truth images without the blur. Each pair was run through their model that added bounding boxes around subjects, which were later refined by human annotators.
The researchers say their model achieves state-of-the-art performance with respect to dynamic deblurring and leads to better restoration results compared with several baselines.
“By comprehensively fusing the deblurring features from different domains, [our model is] able to reconstruct the image with explicit structure and semantic details,” wrote the paper’s co-authors. “Such a design leads to a unified, human-aware, and attentive deblurring network. By explicitly and separately modeling the human-related and [background] blurs, our method can better capture the diverse motion patterns and rich semantics of humans, leading to better deblurring results for both humans and [backgrounds].”