[2022 Spring] ECCV 2016 PlaNet [Eng] (20226186)

[MaheZ20Kaist]

References to related work are missing
Some spelling mistakes ie. technique (check related work and section 5)
The dataset section may be included in Method
A typo/syntactic error in the first line of section 5 after the heading heading "Use of lstm network to classify a set of pictures"

[stitsyuk]

Reviewer: Artyom Stitsyuk 20218256

Actually, I really liked the technical part and the content of the paper review. All sections provide relevant information, which is logically connected. Furthermore, I liked the last “Improvements technics” section, which analyzes possible interesting enhancements such as LSTM network, addition of scene recognition, cross-view image localization and so on. The last idea of detecting and analyzing numbers of cars and names of buildings and stores is a great approach that is able to improve the proposed algorithm even more. Nevertheless, there are many grammatical mistakes in the text and typos that make it difficult to read the paper smoothly. Although the content of the paper review is great and all aspects were described well, I believe that the design of the review and text structure plays a huge role and should be written in user-friendly manner.

I will list the noticed issues point by point:
• In the first statement, the name of the game “Geoguesser” is written with a typo since even in the image it is written differently (“Geoguessr”).
• Typos in writing quotes. I will list some examples:
o “... come directly in mind: «How good a computer can be at« Gueoguesser » ? ...” – the closing quotes are missed.
o “... PlaNet model, which serve as the «benchmark « in the field ...” – two opening quotes.
• Grammar mistakes. I will mention only one example, but there are a lot in the text. I suggest the author to re-read the text again and to correct the typos. For example: “IM2GPS extracts features from the image and the used uses a KNN model to match the prediction with some of the 6 million images stores stored in the model.”
• The captions just after the images. The name of every image is set as “drawing” and is depicted in caption of every image. In my opinion, it should be enumerated and named differently, for example “Fig. 1” and so on as it is done in the text just after the image caption.
• The most important notice, the paper review seems to be cut in the “5. Improvement technics” section just after the figure 15. I checked the github code and noticed that the reason is that the author missed the closing tag: “fig 15: Store front detection using YOLOv3/em>”. Therefore, all text after the image including conclusion is not reflected in the gitbook. I strongly suggest the author to fix this issue ASAP so that other reviewers will be able to review the work properly.

I am sorry again to pay so much attention to the grammar and structure, but I stress it again, the reason is that the technical content is good. Please, just do some corrections in grammar and the review would be great and smooth. Thank you for your work.

[Alex-C137E]

Thank for the comments:
I didn't check how the paper was render from the link the professor gave, so I didn't see the issue with last images and the "drawing" names, because there is not this problem on the GitHub site. I'm going to also improved the redaction and correct all the typos, I have some little issues with that. Sorry.