Auxiliary Equipment Detection in Marine Engine Rooms Based on Deep Learning Model
Round 1
Reviewer 1 Report
This paper proposes a deep learning method for detection and classification of equipment (valves, engines, reservoirs, coolers, etc... ) in a ship's engine room. The paper proposes a series of modifications to a standard RetinaNet neural network that cope well with real time applications (by employing a RepVGG feature extraction network), variable image scales (by employing a Neighbour Erasing and Transferring Mechanism), and a cluttered environment (by an appropriate choice of the loss function). The network was trained with a sufficiently large data set and exhibits promising results. The paper is well written and the results are interesting. There are only two points that were less clear that I would like the authors to clarify before an eventual publication:
What was the data augmentation process used?
Why are the reservoirs and meters more difficult to classify than the other elements?
There is something wrong with equation (1).
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The work presented a solid idea on the subject and a concise overview of the efforts towards this direction. The paper is in very good shape. Abstract and conclusions are descriptive and information provision is easy to be followed.
Some suggestions for further improvement would be:
- Figures 5, 6, and 7 are difficult to be read. Authors could improve the equality of the pictures.
- Figure 8 could be replaced by a similar table.
- Authors could consider refining Figure 9 using homocentric circles. Different processes with the same color could be inside the same circle.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf