Medical Gesture Recognition Method Based on Improved Lightweight Network

Round 1

Reviewer 1 Report

 

The Authors propose three improvements of the known MobileNetv2: (1) including  an Efficient Channel Attention (ECA) module to enhance the network’s feature refining ability, (2) introducing a new activation function R6-SELU to enhance network feature extraction ability, (3) adjusting the width multiplier and resolution multiplier of the model SELU.

Experiments performed with two datasets show that the improved model achieves a good balance between recognition accuracy and model complexity.

The paper is interesting and the results are convincing. To improve the paper I suggest adding comments to the following questions.

(i)             surgeons usually work in gloves,

(ii)           their fingers may be bloody,

(iii)         at least one of their hands may be occupied  (e.g., holding a tool).

 

Did you consider such situations in your datasets?

 

The second important problem is only partially related with your paper:

(a)   how the robot recognizes the start of the doctor’s gesture?

(b) does the robot hand the tool to the surgeon or it puts the tool on the table?

Minor remarks:

Line 98: Relate works,

Formula (3): result of the quotient is unclear,

Line 345: %?,

Line 350: please remark that the difference between two percentages is percent point not percent.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

In the paper, the authors proposed a gesture recognition method applied to a medical assistant robot to collaborate with surgeons to complete surgeries. They propose an improved lightweight convolutional neural network E-MobileNetv2 and The ECA module is added to the original MobileNetv2 network. The experiment showed improvement in accuracy and recognition speed. The following comments are provided:

1 There are some grammar errors and template errors which need corrections.

2 Both part 0 and part 1 are introduction which needs modification.

3 More subparagraphs are needed in part 1 introduction.

4 The meaning of D_f, D_k, M, N need to be explained.

5 Are the experiment in Table 4 in the same experiment settings like the ratio between the training set and the testing set, batch size, etc.?

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The manuscript is improved and the issues are solved after revision.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.