Robust head pose estimation using Dirichlet-tree distribution enhanced random forests

Liu, Yuanyuan; Chen, Jingying; Su, Zhiming; Luo, Zhenzhen; Luo, Nan; Liu, Leyuan; Zhang, Kun

doi:10.1016/j.neucom.2015.03.096

Cited by 27 publications

(5 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Early approaches used classical machine learning models such as Support Vector Regressor (SVR) [105], Localized Gradient Histograms (LCH) [142] or Random Forest (RF) [46,56].…”

Section: Non-linear Regression Methodsmentioning

confidence: 99%

Deep Learning for Head Pose Estimation: A Survey

Asperti

Filippini

2023

SN COMPUT. SCI.

View full text Add to dashboard Cite

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.

show abstract

“…Early approaches used classical machine learning models such as Support Vector Regressor (SVR) [105], Localized Gradient Histograms (LCH) [142] or Random Forest (RF) [46,56].…”

Section: Non-linear Regression Methodsmentioning

confidence: 99%

Deep Learning for Head Pose Estimation: A Survey

Asperti

Filippini

2023

SN COMPUT. SCI.

View full text Add to dashboard Cite

show abstract

“…Experimental results show that this approach can handle depth images with partial occlusions and facial expressions. Inspired by the work of Fanelli et al [ 9 ], Liu et al [ 10 ] proposed Dirichlet-tree distribution enhanced random forests to estimate head pose from RGB images captured in an unconstrained environment. This approach treats head-pose estimation as a supervised classification problem, and reaches an average accuracy rate of 76.2% with 25 head-pose classes on Pointing’04.…”

Section: Related Workmentioning

confidence: 99%

“…For example, in the application of students’ learning-state analysis [ 3 ], inaccurate head-pose estimations often lead to incorrect learning states. Hence, from the classical AdaBoost-based [ 8 ] and random-forests-based [ 9 , 10 ] methods to the current deep neural networks [ 11 , 12 , 13 , 14 ], hundreds of methods have been proposed to pursuit more accurate head-pose estimations.…”

Section: Introductionmentioning

confidence: 99%

“…The most recent mainstream methods [ 9 , 10 , 14 , 15 ] treat head-pose estimation as a supervised classification/regression problem, and need to train models on large-scale datasets with ground-truth head-pose labels. As it is often assumed that a person’s head is a rigid object with three degrees of freedom, the label of head pose is characterized by the pitch, yaw and roll angles with respect to the image plane [ 16 ].…”

Section: Introductionmentioning

confidence: 99%

“…Hence, besides the Pointing’04 dataset, inaccurate labels are common in other head-pose estimation datasets. Although many works [ 10 , 14 ] have reported promising head-pose estimation performance, it is worth noting that these methods are trained and evaluated on datasets with inaccurate or even wrong labels. The methods trained and tested on the same dataset with inaccurate labels tend to create an illusion that their models can accurately estimate head pose.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Head Pose Estimation through Keypoints Matching between Reconstructed 3D Face Model and 2D Image

Liu

Huo

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

Mainstream methods treat head pose estimation as a supervised classification/regression problem, whose performance heavily depends on the accuracy of ground-truth labels of training data. However, it is rather difficult to obtain accurate head pose labels in practice, due to the lack of effective equipment and reasonable approaches for head pose labeling. In this paper, we propose a method which does not need to be trained with head pose labels, but matches the keypoints between a reconstructed 3D face model and the 2D input image, for head pose estimation. The proposed head pose estimation method consists of two components: the 3D face reconstruction and the 3D–2D matching keypoints. At the 3D face reconstruction phase, a personalized 3D face model is reconstructed from the input head image using convolutional neural networks, which are jointly optimized by an asymmetric Euclidean loss and a keypoint loss. At the 3D–2D keypoints matching phase, an iterative optimization algorithm is proposed to match the keypoints between the reconstructed 3D face model and the 2D input image efficiently under the constraint of perspective transformation. The proposed method is extensively evaluated on five widely used head pose estimation datasets, including Pointing’04, BIWI, AFLW2000, Multi-PIE, and Pandora. The experimental results demonstrate that the proposed method achieves excellent cross-dataset performance and surpasses most of the existing state-of-the-art approaches, with average MAEs of 4.78∘ on Pointing’04, 6.83∘ on BIWI, 7.05∘ on AFLW2000, 5.47∘ on Multi-PIE, and 5.06∘ on Pandora, although the model of the proposed method is not trained on any of these five datasets.

show abstract