EfficientPose: Scalable single-person pose estimation

Abstract

Single-person human pose estimation facilitates markerless movement analysis in sports, as well as in clinical applications. Still, state-of-the-art models for human pose estimation generally do not meet the requirements of real-life applications. The proliferation of deep learning techniques has resulted in the development of many advanced approaches. However, with the progresses in the field, more complex and inefficient models have also been introduced, which have caused tremendous increases in computational demands. To cope with these complexity and inefficiency challenges, we propose a novel convolutional neural network architecture, called EfficientPose, which exploits recently proposed EfficientNets in order to deliver efficient and scalable single-person pose estimation. EfficientPose is a family of models harnessing an effective multi-scale feature extractor and computationally efficient detection blocks using mobile inverted bottleneck convolutions, while at the same time ensuring that the precision of the pose configurations is still improved. Due to its low complexity and efficiency, EfficientPose enables real-world applications on edge devices by limiting the memory footprint and computational cost. The results from our experiments, using the challenging MPII single-person benchmark, show that the proposed EfficientPose models substantially outperform the widely-used OpenPose model both in terms of accuracy and computational efficiency. In particular, our top-performing model achieves state-of-the-art accuracy on single-person MPII, with low-complexity ConvNets.

In-Motion-App for remote General Movement Assessment: a multi-site observational study

“…Third, the performance of the infant motion tracker is reported as percentage of points within a circular area centred at the manually annotated body point for the 5493 frames. In accordance with the established metric for evaluating pose-estimation, 17 radius of the circular area was set to 10% of the infant head size and was normalised to adjust for different scaling (ie, video zoom) ( figure 2 ).…”

“…In accordance with the established metric for evaluating pose-estimation, 17 radius of the circular area was set to 10% of the infant head size and was normalised to adjust for different scaling (ie, video zoom) (figure 2).…”

A Pose-Based Feature Fusion and Classification Framework for the Early Prediction of Cerebral Palsy in Infants

“…Research into the feasibility of using pose-based assessment for GMA is ongoing, with several papers contributing advances in this area. In [10], [18], [27], [30] and [36], pose estimation and domain adaptation methods for video sequences of infants are proposed, with each suggesting that their methods are a step towards automated posebased GMA. However, each typically assesses the effectiveness of the extracted pose rather than the effect this might have upon final classification.…”

Performance Analysis in Ski Jumping with a Differential Global Navigation Satellite System and Video-Based Pose Estimation

“…The PosEst system estimated the x and y coordinates of 16 body parts (i.e., head top, upper neck, shoulders, elbows, wrists, upper chest, right/mid/left pelvis, knees, and ankles) in the video frames in a frame-by-frame manner using a state-of-the-art convolutional neural network (ConvNet) for high-precision pose estimation. For further technical details about ConvNet, the reader is referred to [ 41 ]. To perform pose estimations of the ski jumpers, the PosEst was trained, validated and tested on 5064 randomly selected video frames (i.e., 3686 (73%) for training, 365 (7%) for validation, and 1013 (20%) for testing) from an internal database of jumps recorded between January 2014 and June 2019.…”