Semantic Visual SLAM Algorithm Based on Improved DeepLabV3+ Model and LK Optical Flow

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. In abstract,  refine/rewrite  the statement with facts and figures such as accuracy, highest performance etc. " The experimental results show that the proposed semantic visual SLAM algorithm can effectively reduce the influence of dynamic targets on the system, has a higher localization accuracy, and compared with other advanced algorithms, such as DynaSLAM, has the highest performance in indoor dynamic environments while considering both localization accuracy and real-time performance." 

2. Suggested to do space and time complexity of the proposed algorithms.

3. Include the camera specification which you have used in the research.

4. Justify it the 'To further improve the localization accuracy of visual SLAM algorithms for AGVs in  dynamic environments'.

5. Clearly mention the feature extraction, feature matching, motion estimation, pose optimization, state update, and keyframe selection.

6. suggested to mention the reference for equations which were used in the paper.

7. In Eqn 7, what is the value of judgement threshold?

8. Suggested to estimate the   time to remove the static keyframes with dynamic disturbance, the background region occluded by the dynamic target is repaired.

9. Suggested to estimate the time to Octree Map Construction?

10. Show the calculation for 89% etc.?

11, Suggested to estimate the time to Dense point cloud map construction.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The provided manuscript describes a novel algorithm applied to the visual localization and mapping for a mobile robot. The manuscript is extremely well written, with clear descriptions of the methodology and the results are described in depth with comparison to equivalent algorithms.

There are few small improvements I would advise:

1. Authors should address the relatively low accuracy of backbone models in Table 1. While this obviously does not affect the performance, authors should address how a final model would behave with a more accurate, or less accurate, backbone.

2. In conclusions, authors should explain the limitations of the manuscript in a more detailed manner.

Due to the above, I reccommend minor revisions to the manuscript prior to publication.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf