Action Recognition of Taekwondo Unit Actions Using Action Images Constructed with Time-Warped Motion Profiles

Round 1

Reviewer 1 Report

Contribution/Summary: The study introduces a novel action recognition model tailored for Taekwondo unit actions, utilizing joint motion data acquired via wearable inertial measurement unit (IMU) sensors. The model synthesizes action images by mapping joint motion profiles onto the RGB color space, encapsulating the motion dynamics of a single unit action within a solitary image. A time-warping technique is applied to adjust motion profiles in relation to the velocity of the action, enhancing the representation of rapid movements within the action images.

Comments/Suggestions:

1. Provide more details about the dataset compiled from 40 Taekwondo experts, such as the demographics of the participants and the specific actions performed, to enhance the reproducibility of the study.

2. Include a comparison with existing action recognition models for Taekwondo or other martial arts to highlight the novelty and superiority of the proposed model.

3. Discuss the limitations of the proposed model, such as its performance in real-world scenarios with varying environmental conditions and the potential impact of different attire worn by competitors.

4. Conduct a user study or expert evaluation to assess the practical usability and effectiveness of the proposed model in a real Taekwondo competition setting.

5. Provide a detailed explanation of the time-warping technique used to adjust motion profiles, including the specific equations and algorithms employed, to enhance the clarity and reproducibility of the methodology.

6. Discuss the potential applications and implications of the proposed model beyond action recognition in Taekwondo, such as in other sports or physical rehabilitation settings.

7. Include a discussion on the computational efficiency and resource requirements of the proposed model, considering the potential deployment on wearable devices or real-time action recognition systems.

8. Consider conducting a sensitivity analysis or robustness testing to evaluate the performance of the proposed model under different parameter settings or variations in the input data.

9. The authors are invited to include some recent references, especially those related to  Deep Convolutional Neural Networks. 

10. For instance, the authors may include the following interesting references (and others):

a. https://www.mdpi.com/2073-431X/12/8/151

b. https://www.taylorfrancis.com/chapters/edit/10.1201/9781003393030-10/learning-modeling-technique-convolution-neural-networks-online-education-fahad-alahmari-arshi-naim-hamed-alqa

Can be improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper proposes a novel approach to action recognition within the sport of Taekwondo using a time-warping technique on inertial data collected from wearable sensors. The authors demonstrate promising results of 0.998 accuracy and similar in precision, recall and F1 scores. Overall, it is a very interesting and well presented paper!

Some minor comments:

1. One limitation is that the authors outline limitations in current vision based strategies but do not provide an exhaustive review of the area. This may lead to questions regarding the models effectiveness, comparatively against the state-of-the-art.

2. The experiments are limited to 40 experts in the sport. Can this be considered fully representative of all experts across the sport? Also, what was the make-up of the participants in terms of age/gender/weight? Was it balanced? If not, how would this affect the model? There may also be other factors to consider. The work would benefit if theses questions were addressed in the text.

In summary, the paper is well-written and offers a compelling direction for action recognition in the sport of Taekwondo using advanced sensor data and neural network architectures. The paper may benefit from an expanded discussion on the limitations, potential biases in the dataset, and comparative analysis with other methods.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors considered my comments and suggestions. Good luck.

A final proofread would be useful