Improving Temporal Interpolation of Head and Body Pose using Gaussian Process Regression in a Matrix Completion Setting
This work addresses the challenge of pose estimation in conversational group interactions for researchers in computer vision and multimodal sensing, though it is incremental as it builds on existing matrix completion methods.
The paper tackles the problem of head and body pose estimation with temporally sparse labeled data by proposing a model that replaces Laplacian smoothing with Gaussian process regression, incorporates head-body coupling, and uses nuclear norm minimization in a matrix completion setting. It outperforms the state-of-the-art on the SALSA dataset, achieving approximately 62% head pose accuracy and 70% body pose accuracy with only 5% ground truth labels as training data.
This paper presents a model for head and body pose estimation (HBPE) when labelled samples are highly sparse. The current state-of-the-art multimodal approach to HBPE utilizes the matrix completion method in a transductive setting to predict pose labels for unobserved samples. Based on this approach, the proposed method tackles HBPE when manually annotated ground truth labels are temporally sparse. We posit that the current state of the art approach oversimplifies the temporal sparsity assumption by using Laplacian smoothing. Our final solution uses: i) Gaussian process regression in place of Laplacian smoothing, ii) head and body coupling, and iii) nuclear norm minimization in the matrix completion setting. The model is applied to the challenging SALSA dataset for benchmark against the state-of-the-art method. Our presented formulation outperforms the state-of-the-art significantly in this particular setting, e.g. at 5% ground truth labels as training data, head pose accuracy and body pose accuracy is approximately 62% and 70%, respectively. As well as fitting a more flexible model to missing labels in time, we posit that our approach also loosens the head and body coupling constraint, allowing for a more expressive model of the head and body pose typically seen during conversational interaction in groups. This provides a new baseline to improve upon for future integration of multimodal sensor data for the purpose of HBPE.