Deep Regression Forests for Age Estimation

Abstract

Age estimation from facial images is typically cast as a nonlinear regression problem. The main challenge of this problem is the facial feature space w.r.t. ages is heterogeneous, due to the large variation in facial appearance across different persons of the same age and the nonstationary property of aging patterns. In this paper, we propose Deep Regression Forests (DRFs), an end-to-end model, for age estimation. DRFs connect the split nodes to a fully connected layer of a convolutional neural network (CNN) and deal with heterogeneous data by jointly learning input-dependant data partitions at the split nodes and data abstractions at the leaf nodes. This joint learning follows an alternating strategy: First, by fixing the leaf nodes, the split nodes as well as the CNN parameters are optimized by Back-propagation; Then, by fixing the split nodes, the leaf nodes are optimized by iterating a step-size free and fastconverging update rule derived from Variational Bounding. We verify the proposed DRFs on three standard age estimation benchmarks and achieve state-of-the-art results on all of them.

Deep Differentiable Random Forests for Age Estimation

“…This paper summarizes two of our preliminary works [28], [29] into a unified optimization framework, i.e., alternatively learning split nodes by Back-propagation and learning leaf nodes by Variational Bounding and has following extensions: First, we introduce two methodological improvements, i.e., the two Deterministic Annealing processes introduced into the learning of split and leaf nodes, respectively, to avoid poor local optima and obtain better estimates of tree parameters free of initial parameter values. Second, we provide more experimental results and discussions, such as ablation experiments to study the influence of different designs and variants of our methods and updated state-ofthe-art results on the three age estimation datasets.…”

“…35, R(π, Θ; S) may converge to a poor local minimum, if µ (0) and σ (0) are not well initialized. In our previous work [29], we did kmeans clustering on {y i } N i=1 to obtain |L| subsets, then initialized µ (0) and σ (0) according to cluster assignment. Here, inspired by [23], [24] we propose a deterministic annealing algorithm for the above optimization problem, which leads to an initialization free solution to avoid poor local minimum.…”

Ordinal distribution regression for gait-based age estimation

“…The results of MAE and CS (k = 5) with three different networks are shown in Ta Compared with a simple CNN, GL-CNN achieves a better performance in age estimation. It can be seen from Table 2 that 1) compared with CNN and VGG16, GL-CNN achieves the best performance in two criteria; 2) although VGG16 has more parameters, GL-CNN effectively learns more detail information through combining a global and three local structures, resulting in an improved perfor-Young 3 (3) 4 (4) 4 (5) 7 (8) 8 (9) 10 (9) 11 (10) 14 (15) 17 (16) 21 (21) 23 (24) 25 (24) 26 (26) 27 (26) 29 (30) 31 (32) 36 (37) 38 (41) Figure 6. Examples of gait-based age estimation results by the proposed approach on OULP-Age dataset.…”

“…It indicates that ODL is more effective in learning the ordinal relationship among different age than a single cross-entropy loss. Methods MAE CS (k = 5) OR-CNN [22] 3.27 73.0%* DEX [27] 3.25 N/A Ranking-CNN [3] 2.96 85.0%* VGG16 + Mean-Variance [23] 2.41 90.0%* DRFs [32] 2.17 91.3% VGG16 + ODL(λ = 0) 2.30 91.1% VGG16 + ODL(λ = 0.25) (Ours) 2.16 92.9% Table 5. Comparisons between our approach and the state-of-theart methods on the MORPH Album II dataset in terms of MAE and CS value (*: the value is read from the reported CS curve).…”

“…Human age estimation remains to be an active research topic, which plays an important role on many potential applications, such as video surveillance, social networking, human-computer interaction, etc. Existing human age estimation methods are mainly based on facial images [2,3,22,23,31], which are very informative and easy to be estimated. The performance of the face-based age estimation approaches, however, will compromise when the face is occluded, for example, with sunglass or makeup.…”