A discussion on the validation tests employed to compare human action recognition methods using the MSR Action3D dataset

By José Ramón Padilla-López, Alexandros André Chaaraoui, Francisco Flórez-Revuelta. 2014

47 citations recorded by the source corpus.

This paper aims to determine which is the best human action recognition method based on features extracted from RGB-D devices, such as the Microsoft Kinect. A review of all the papers that make reference to MSR Action3D, the most used dataset that includes depth information acquired from a RGB-D device, has been performed. We found that the validation method used by each work differs from the others. So, a direct comparison among works cannot be made. However, almost all the works present their results comparing them without taking into account this issue. Therefore, we present different rankings according to the methodology used for the validation in orden to clarify the existing confusion.

Subfields: Artificial Intelligence, Biomedical Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Fei Han
2. Liliana Lo Presti
3. Marco La Cascia
4. Hao Zhang
5. Alan L. Yuille
6. Yizhou Wang
7. William A. Hoff
8. Brian Reily

In Biomedical Engineering

1. Ennio Gambi
2. Susanna Spinsante
3. Enea Cippitelli
4. Fei Han
5. William A. Hoff
6. Brian Reily
7. Alan L. Yuille
8. Liliana Lo Presti

In Computer Vision and Pattern Recognition

1. Fei Han
2. Hao Zhang
3. Jun Liu
4. William A. Hoff
5. Brian Reily
6. Liliana Lo Presti
7. Marco La Cascia
8. Yizhou Wang

Citations in the corpus

• Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, methodology): “A large number of approaches used this dataset for evaluation and validation [127].”
• Mining 3D Key-Pose-Motifs for Action Recognition (2016, background): “However, it is shown in [15] that the way how the data are split (i.”
• A Human Activity Recognition System Using Skeleton Data from RGBD Sensors (2016, methodology): “[49] summarized all the validation methods for this dataset and recommend using all the possible combinations of 55 subjects splitting or using LOAO.”
• 3D skeleton-based human action classification: A survey (2016, methodology): “It is already known that the choice of the validation protocol affects the reported accuracy values on the MSRA-3D dataset [149].”