*Authors, reviewers, and meta-reviewers are advised that automated emails from the conference management system (CMT) may be filtered as spam; it is the responsibility of all participants to regularly check spam folders/whitelist „msr-cmt.org“ and ensure such messages are received.*

Don’t reply to emails from CMT. If you have any issues, contact the program chairs directly ([email protected])

Thank you for serving as a reviewer or meta-reviewer!

Your work is essential to ICDAR 2026

Outstanding reviewers will be recognized during ICDAR 2026

Meta-Reviewer (Senior PC Member) instructions for CMT (taken from ICDAR2025)

(Info adapted from Guidelines ICDAR 2025)

General Advice

Be specific and detailed. Do not simply give summary judgments (“not novel”, “unclear”, “incorrect”) – justify them in detail. Your comments will be more helpful to the Meta-Reviewer (Senior PC-Member) and the authors than your scores
Be professional and courteous. Belittling, sarcastic, or overly harsh remarks have no place in the reviewing process.
Be careful about giving away your identity by asking the authors to cite only your own papers and no other papers.
Be participative. Discuss with the other reviewers once all reviews are in. Read the rebuttal and discuss any outstanding issues with the rest of the reviewers and the Meta-Reviewer (Senior PC-Member).
Be on time and responsive.
The use of LLMs to write your reviews is strictly forbidden.
Proofread and spellcheck your reviews. You are allowed to use an LLM to check short phrases for clarity. This includes using grammar checkers such as Grammarly.

Reviews should be concrete, constructive and clear

Reviewer comments should be informed by your reading of the paper, should clearly identify the paper’s merits and limitations, and make clear where any issues are found along with specific actions to address those issues where possible.
Reviews with generic complaints that lack clear ties to specific sections of the paper are unhelpful. Concrete steps to address limitations both improve authors’ understanding of their work and improve the quality of papers accepted for publication.
Aim to provide constructive criticism in a manner that you would like to receive yourself, where the reviewer’s comments can be both easily understood and acted upon.

Suggestions for missing material / additional work

It is appropriate to identify important omissions from a paper. This might include prior work that should have been discussed or additional experiments that should have been performed.
Keep in mind, however, that even though ICDAR 2026 employs an “author response” (rebuttal) phase, this is only to allow authors to voice their disagreement with the factual aspects of a review; it does not allow an author to promise to do additional work to fix problems in a paper.
While authors may certainly update the final version of their paper before the camera-ready deadline if it is accepted, the Program Chairs will make their final acceptance decisions based on the original submission and not on promises to do additional work identified by reviewers.

Confidentiality

You should consider the papers you have been assigned to review as confidential. You must not use them in any way outside of the reviews you are performing for ICDAR 2026. If a paper is ultimately accepted and published, at that point, you can, of course, reference the work in one of your own papers.
If you encounter an issue with one of the papers you are assigned to review (e.g., an integrity problem), you should bring up the matter with the Program Co-Chairs. To protect confidentiality, you should not discuss such matters with anyone else, nor should you share the papers you have been assigned to review with others to ask their opinions.
Inputting an ICDAR submission to an LLM can be considered a violation of this confidentiality requirement, which is another reason why you must avoid using LLMs during reviewing, except to check for minor grammar mistakes in your review.

Anonymity

ICDAR 2026 employs a double-blind review process. This means authors should not identify themselves in their papers, and reviewers should not identify themselves in their reviews. Because there is no perfect way to guarantee anonymity, this also means reviewers should not try to figure out who wrote a given paper.
The PC chairs have already checked papers for obvious anonymity issues. In the rare scenario that you encounter anonymization issues in a paper you are reviewing, bring this issue to the attention of the Meta-Reviewer (Senior PC-Member), but continue reviewing as normal, unless you are instructed to do otherwise. Remember that it is not the reviewers‘ responsibility to desk-reject a paper.
If an author has inadvertently made identification possible, reviewers are instructed to do their best to ignore this information when reviewing the paper.
For example, an author might reference a specific research project, or a public repository for data or source code (e.g., GitHub). Reviewers are not required to consider such outside materials, but if they do so, they should avoid attempting to connect them to an identification of the author.
Note also that self-archiving is permitted. For example, posting the paper on Arxiv is acceptable. The same principle as before applies: you, as a reviewer, should do your best not to be conditioned during the review process. Do not actively look for the paper online.

Conflicts of interest

The CMT system we use for ICDAR 2026 has certain automated features to avoid conflicts of interest. But, again, such features are not perfect. They rely on receiving good information from everyone participating in this process (including, for example, domain conflicts between reviewers and authors), as well as external information (e.g., DBLP co-authorship data).
If you detect a conflict of interest in a paper you have been assigned to review, notify the Program Co-Chairs immediately; do not complete a review for the paper in question.

Your Review

General

As an ICDAR 2026 reviewer (Program Committee member), can I delegate reviews assigned to me to someone else I know?

The short answer is „no.“ When we invited leaders in the document analysis research community to serve as PC members, it was with the intention that they would take personal responsibility for all of the papers assigned to them. While the option of delegating reviews was a practice in past conferences, this is no longer the case.

There are good reasons for this change. The integrity of the review process is very important to both the conference itself and the publisher of our proceedings, Springer Nature. Numerous emails were sent to reviewers before the start of the review period, reminding them to fill out their domain conflicts and also their topic areas. This is to avoid conflicts of interest and to help align submissions with qualified reviewers. Springer-Nature requires that we provide affiliation details for everyone involved in helping to decide the ICDAR program, including all of our reviewers. This is because some „bad actors“ are using fake peer review processes, where reviewers are not truly independent and where authors „pay to publish“ low-quality papers. By identifying those involved in determining the program and their affiliations, we provide assurance that ICDAR 2026 is a high-quality conference with a rigorous review process.

Paper summary

Summarize in your own words the paper, its key ideas, its contributions, and their significance.
This summary helps the Meta-Reviewer (Senior PC-Member) and the authors understand the rest of your review and be confident that you understand the paper.

Paper strengths

Detail the strengths of the paper.
These can be, for example, interesting ideas, an insightful organization of related work, new tools, or impressive results.
Be specific: Your comments will be much more helpful to the Meta-Reviewer (Senior PC-Member) and the authors than your score.

Paper weaknesses (if any)

Detail the important weaknesses you identified and justify them. For example:

If you found the method difficult to understand, explain what the authors did wrong and give examples.
If the claimed contributions are not backed up by experiments, explain what should have been done.
Note that we do not expect in-depth comparisons against papers for which source code is not openly available, since the results of such papers cannot be easily and independently reproduced.
Note that we do not expect comparisons against papers that appear on arXiv only (have not gone through a proper peer review process).

You can also note minor issues that are easy to fix in the camera-ready paper. If you do so, please state that you consider them minor. They should not affect your decision about this paper (as there is no guarantee that the authors will actually make these changes). For example:

missing references that would enrich the discussion but do not really change the paper’s novelty. Do not give away your identity by asking the authors to cite only your own papers!
occasional typographic errors;
minor unclear points;

Be specific: Do not simply give summary judgments. Writing “not novel”, “unclear”, “incorrect” is not sufficient! Your comments will be much more helpful to the Meta-Reviewer (Senior PC-Member) and the authors than your score

Justification

Balance strengths and weaknesses to explain why the paper should be accepted or not.

Minor weaknesses are easy to fix for the final version and should not weigh into this decision.
It is not because a paper beats the state of the art that it should be accepted.
It is not because a paper does not beat the state of the art that it should be rejected.
It is not because a method is novel – or claims novelty – that it is necessarily interesting.
It is not because the novelty seems limited that the contribution is not important:

Remember that at this stage of the reviewing process, you can ask the authors to clarify important points in their rebuttal. Be clear about what you would expect to see in the rebuttal.

Final rating and comments (after rebuttal)

Before you make your final decision, please read the other reviews and the authors’ rebuttal carefully.

Take advantage of the discussion period to exchange with other reviewers and the AC about any doubts or points you consider important.

Defend your position with clear arguments during the discussion. Maybe the other reviewers have missed something. Remember that you can change your final rating during this short discussion period if needed.

Ponder your first recommendation. There may be something you misunderstood or overlooked.

Justify your final decision in the final comments. The authors worked hard on their paper and their rebuttal, and simply writing the rebuttal did not convince you is not good enough.

Note that there is no option for borderline in the final ICDAR 2026 rating. At this stage, and after the authors have had a chance to respond, you need to make up your mind about the paper.

Acceptance Criteria

Should this paper be considered for a best paper award?

Tick this box for major advances that will impact the field long term; ideas will be used by many people, create new capabilities, etc.

Accept:

Multiple types of papers fit into this category

Potential to be very significant; worthwhile for the whole community to hear about. Incremental steps that expand the sum of the community’s knowledge or add bricks to the cathedral of knowledge; papers introducing useful tools; papers of interest to a subcommunity.

Creative ideas that are hard to judge but could be promising — no one knows the future, so we should give the benefit of the doubt to plausible ideas.

Reject:

Unlikely to be significant. Works with obvious technical flaws. Nothing new to be learned by someone who already knows the field.

Decision Process

Program Chairs (PCs) assign papers to Meta-Reviewers (Senior PC Member), about 10-15 papers per Meta-Reviewer for ICDAR 2026
Meta-Reviewers and Reviewers bid for papers, with the help of CMT
Papers are assigned to Meta-Reviewers and Reviewers (3 reviewers per paper) using an optimization algorithm that considers bidding, paper load, and conflict constraints
Reviewers submit initial reviews, Meta-Reviewers check the quality of reviews, chase late reviewers, and assign emergency Reviewers as necessary
Authors receive reviews with a mixture of gasps, grimaces, grumbles, and the occasional grin. After much thought and re-reading of the paper and reviews, the authors submit rebuttals
Discussion among Reviewers and Meta-Reviewers, based on all reviews, rebuttal, and paper. Reviewers update their ratings and justification
Meta-Reviewers make decisions and write meta-reviews. The decision and meta-review are recorded by the Meta-Reviewer for each paper and checked/approved by the PCs. Meta-Reviewers can reach out to the PCs to discuss borderline papers. Additional opinions may be sought from other expert Meta-Reviewers after checking conflicts
PCs make final determination and verify decisions, based on the recommendations of the Meta-Reviewers