Head pose estimation (HPE) is an active and popular area of research. Over the years, many approaches have constantly been developed, leading to a progressive improvement in accuracy; nevertheless, head pose estimation remains an open research topic, especially in unconstrained environments. In this paper, we will review the increasing amount of available datasets and the modern methodologies used to estimate orientation, with a special attention to deep learning techniques. We will discuss the evolution of the field by proposing a classification of head pose estimation methods, explaining their advantages and disadvantages, and highlighting the different ways deep learning techniques have been used in the context of HPE. An in-depth performance comparison and discussion is presented at the end of the work. We also highlight the most promising research directions for future investigations on the topic.

In this article we introduce the notion of Split Variational Autoencoder (SVAE), whose output x is obtained as a weighted sum σ ⊙ x1 + (1 − σ) ⊙ x2 of two generated images x1 , x2 , and σ is a learned compositional map. The composing images x1 , x2 , as well as the σ-map are automatically synthesized by the model. The network is trained as a usual Variational Autoencoder with a negative loglikelihood loss between training and reconstructed images. No additional loss is required for x1 , x2 or σ, neither any form of human tuning. The decomposition is nondeterministic, but follows two main schemes, that we may roughly categorize as either "syntactic" or "semantic". In the first case, the map tends to exploit the strong correlation between adjacent pixels, splitting the image in two complementary high frequency sub-images. In the second case, the map typically focuses on the contours of objects, splitting the image in interesting variations of its content, with more marked and distinctive features. In this case, according to empirical observations, the Fréchet Inception Distance (FID) of x1 and x2 is usually lower (hence better) than that of x, that clearly suffers from being the average of the former. In a sense, a SVAE forces the Variational Autoencoder to make choices, in contrast with its intrinsic tendency to average between alternatives with the aim to minimize the reconstruction loss towards a specific sample. According to the FID metric, our technique, tested on typical datasets such as Mnist, Cifar10 and Celeba, allows us to outperform all previous purely variational architectures (not relying on normalization flows).

No abstract