Though existing face hallucination methods achieve great performance on the global region evaluation, most of them cannot recover local attributes accurately, especially when super-resolving a very low-resolution face image from 14 × 12 pixels to its 8 × larger one. In this paper, we propose a brand new Attribute Augmented Convolutional Neural Network (AACNN) to assist face hallucination by exploiting facial attributes. The goal is to augment face hallucination, particularly the local regions, with informative attribute description. More specifically, our method fuses the advantages of both image domain and attribute domain, which significantly assists facial attributes recovery. Extensive experiments demonstrate that our proposed method achieves superior visual quality of hallucination on both local region and global region against the state-of-the-art methods. In addition, our AACNN still improves the performance of hallucination adaptively with partial attribute input.

Face hallucination is a generative task to super-resolve the facial image with low resolution while human perception of face heavily relies on identity information. However, previous face hallucination approaches largely ignore facial identity recovery. This paper proposes Super-Identity Convolutional Neural Network (SICNN) to recover identity information for generating faces closed to the real identity. Specifically, we define a super-identity loss to measure the identity difference between a hallucinated face and its corresponding high-resolution face within the hypersphere identity metric space. However, directly using this loss will lead to a Dynamic Domain Divergence problem, which is caused by the large margin between the high-resolution domain and the hallucination domain. To overcome this challenge, we present a domain-integrated training approach by constructing a robust identity metric for faces from these two domains. Extensive experimental evaluations demonstrate that the proposed SICNN achieves superior visual quality over the state-of-the-art methods on a challenging task to super-resolve 12$\times$14 faces with an 8$\times$ upscaling factor. In addition, SICNN significantly improves the recognizability of ultra-low-resolution faces.

This paper proposes LightBib, a novel marathoner recognition system utilizing visible light communications (VLC). In our system, each runner wears a pair of light strips which transmits the runner's ID with VLC, while cameras set along the running route can be used to decode the runner's ID from the pixels in the captured images. Our transmitter utilizes a frequency shift keying modulation, sending the information with an on-off pattern at a certain frequency. Commodity rolling shutter cameras, such as consumer pocket cameras and smartphone cameras, can be used to receive the transmission. The received images can then be post-processed, where image areas corresponding to a certain runner can be detected and recognized by extracting the ID embedded in the pixels with very high accuracy. This is useful in efficiently and accurately locate the video segments with a particular runner out of a large collection of video files. Our experimental results indicate that the average recall to identify each of the 5 runners in a single image is close to 90%.