Soft-ranking Label Encoding for Robust Facial Age Estimation

Abstract

Automatic facial age estimation can be used in a wide range of real-world applications. However, this process is challenging due to the randomness and slowness of the aging process. Accordingly, in this paper, we propose a comprehensive framework aimed at overcoming the challenges associated with facial age estimation. First, we propose a novel age encoding method, referred to as Soft-ranking, which encodes two important properties of facial age, i.e., the ordinal property and the correlation between adjacent ages. Therefore, Soft-ranking provides a richer supervision signal for training deep models. Moreover, we also carefully analyze existing evaluation protocols for age estimation, finding that the overlap in identity between the training and testing sets affects the relative performance of different age encoding methods. Finally, since existing face databases for age estimation are generally small, deep models tend to suffer from an overfitting problem. To address this issue, we propose a novel regularization strategy to encourage deep models to learn more robust features from facial parts for age estimation purposes. Extensive experiments indicate that the proposed techniques improve the age estimation performance; moreover, we achieve state-of-the-art performance on the three most popular age databases, i.e., Morph II, CLAP2015, and CLAP2016.

Empowering Things With Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial Intelligence of Things

“…9) Crowd Counting: In the video surveillance scenario, it is necessary to count the crowd in both indoor and outdoor areas and prevent crowd congestion and accident. For practical AIoT applications with crowd counting ability, WI-FI, Bluetooth, The processed result by using different perceiving methods, i.e., semantic segmentation [85]; object detection [106]; text spotting [91]; human parsing [86]; human pose estimation [107]; face detection, alignment, and facial attribute analysis [96], [108]; depth estimation [109]. and camera-based solutions have been proposed by estimating the connections between smartphones and WI-FI access points or Bluetooth beacons [126] or estimating the crowd density of a crowd image [127].…”

“…(a) A frame from the video "Walking Next to People" 11 . (b)The processed result by using different perceiving methods, i.e., semantic segmentation[85]; object detection[106]; text spotting[91]; human parsing[86]; human pose estimation[107]; face detection, alignment, and facial attribute analysis[96],[108]; depth estimation[109].…”

Deep Ordinal Regression with Label Diversity

“…Others have observed that it is easier for a human to distinguish differences in age between two persons, rather than their absolute age and used this as a design principle for ordinal regression [7], [20]. Alternative methods have focused on various soft encodings of the age over classes, where the elements of the probability vector are proportional to the distance from the true class [9], [21], [22]. In this way both the ordinal and metric information can be effectively encoded in the labels.…”

Self-Paced Deep Regression Forests for Facial Age Estimation

“…This problem has been studied in recent works [10,11] as how to learn robust and discriminative facial features through a rather deep neural network with a regularized loss function. However, they do not truly distinguish the confusing and noisy samples from regular ones, and yet not exclude the interference caused by the unreliable samples, resulting in unsatisfactory robustness.…”