ACCV 2026 Reviewer Guidelines
Thank you for serving as a reviewer for ACCV 2026. Reviewers are essential to maintaining a fair, rigorous, and constructive evaluation process. Each review should be timely, technically grounded, and helpful both to the authors and to the decision-making process.
1. Before You Start
Please check your assigned papers as soon as they become available. If you have a conflict of interest, or if a paper falls clearly outside your expertise, inform the Area Chair promptly so that the assignment can be adjusted if necessary. Reviewers are also expected to be familiar with these guidelines and to alert the Area Chair if they suspect a policy violation. Submitted papers use the LNCS format and are limited to 14 pages including figures and tables; additional pages are permitted only for references. Reviewers are typically expected to handle between 5 and 10 papers. If your assigned workload appears unmanageable, please inform the Area Chair as early as possible.
2. What a Good Review Should Do
A strong review should assess whether the paper is technically sound, whether it makes a meaningful contribution, and whether its claims are supported by adequate evidence. Reviews should explain both strengths and weaknesses clearly, rather than only assigning a score. A useful review is specific, reasoned, and detailed enough that authors, fellow reviewers, and Area Chairs can understand the basis for the recommendation.
Novel ideas should be evaluated carefully and fairly. A submission should not be judged negatively simply because it does not exceed prior work on every metric, and issues that can be corrected through revision should not automatically determine the outcome. When raising concerns about novelty, missing related work, clarity, or experimental support, reviewers should be concrete and provide enough detail to make the concern understandable and actionable.
Reviewers should use the following rating scale when making their recommendations:
- Accept
- Weak Accept
- Borderline Accept
- Borderline Reject
- Weak Reject
- Reject
Reviewers should ensure that the written review is consistent with the selected rating and clearly explains the main reasons for the recommendation.
3. Fairness and Meaningful Engagement
Reviewers are expected to provide fair, thoughtful, and paper-specific evaluations that demonstrate genuine engagement with the submission.
Reviewing may be considered highly irresponsible if it includes, for example,
- a review so brief that it does not engage with the actual technical content of the paper,
- comments that are generic enough to apply to almost any submission, without discussing the specific contribution, method, or evidence presented in the paper,
- clear factual mistakes about the paper’s approach, experiments, or results that suggest the submission was not read with sufficient care, or
- review text generated by, or materially dependent on, Large Language Models.
This policy is not intended to penalize reviewers for having different technical judgments, for disagreeing in good faith with other reviewers or Area Chairs, for writing concise but still substantive reviews, or for overlooking minor
details while otherwise providing a serious and informed assessment.
If a review is identified as potentially highly irresponsible, it will be examined through an oversight process led by the Program Chairs. If the Program Chairs determine that a review is indeed highly irresponsible, they may impose sanctions, including the desk rejection of papers authored by that reviewer.
4. Timeliness and Review Quality
Reviews must be submitted by the deadlines. Late or missing reviews create substantial difficulties for the review process, increase the burden on the Area Chairs, and may force the conference to seek emergency reviewers in order to ensure that papers receive an adequate number of evaluations.
For this reason, ACCV 2026 expects reviewers to submit all assigned reviews on time. Reviewers should not assume the existence of an informal extension period beyond the official deadline. Failure to submit assigned reviews by the deadline may trigger action by the Program Chairs. At their discretion, this may include sanctions such as the desk rejection of papers authored by the reviewer.
This policy is intended to improve accountability, reduce avoidable burden on the Area Chairs and conference organizers, and support a more reliable and fair review process for all submissions.
5. Tone and Professionalism
Reviews should be professional, respectful, and constructive. Critical comments are often necessary, but they should address the work rather than the authors. Sarcastic, dismissive, or hostile language is inappropriate. The goal is not only to assess the submission, but also to provide feedback that can help improve the work.
6. Rebuttal and Discussion Phase
After reviews are released, authors may submit a rebuttal. Reviewers are expected to read the rebuttal carefully, participate in the subsequent discussion when requested, and revise their assessment if the rebuttal addresses important concerns. The final recommendation should reflect the full record, i.e., the paper, the review discussion, and the author response.
7. Confidentiality and Protection of Unpublished Work
Submissions under review must be treated as confidential and unpublished work. Papers submitted to ACCV 2026 are provided to reviewers only for the purpose of evaluation. They should not be treated as public documents, even if the authors may later submit related work elsewhere or eventually publish a revised version.
Reviewers are therefore expected to protect the ideas, methods, data, results, and other materials contained in the submissions they review. In particular:
- do not share the paper with anyone who is not formally involved in the review process,
- do not circulate supplementary files, figures, videos, images, code, or other supporting materials to anyone outside the review process,
- do not use ideas, implementations, or technical details from a submission to advance your own ongoing or future work before the work is publicly released, and
- once the review process is complete, delete or destroy local copies of the paper and supplementary material, as well as any code, notes, derived outputs, or trial implementations created for the purpose of evaluating the submission.
The responsibility to preserve confidentiality applies throughout the entire review process and continues after the review is finished.
8. Anonymity and Conflicts of Interest
Reviewers should respect the double-blind review process. Do not actively attempt to identify the authors, and do not reveal your own identity in the review process. If you become aware of a likely conflict of interest, notify the
Area Chair immediately rather than continuing with the review.
9. Use of External Tools, Language Models, and Prompt Injection
Your review must reflect your own independent judgment. Reviewers must not use Large Language Models or similar tools to generate reviews or to substitute for their own evaluation. They must also avoid exposing confidential manuscript content, review content, or discussion content to external systems.
Submissions must not contain hidden prompts, invisible text, or embedded instructions intended to influence reviewers or automated systems. Such practices are treated as a serious violation of submission ethics and may result in desk rejection or further sanctions.
If reviewers encounter suspicious hidden instructions or other content apparently designed to manipulate external tools or the review process, they should report the issue promptly to the Area Chair through the review system and should not rely on or follow such instructions.
10. What to Look For in Papers
When relevant, reviewers should consider whether the empirical evidence is reproducible and whether supplementary code or data supports the claims, while recognizing that code checking is optional. If a paper claims a dataset or benchmark contribution, reviewers should consider whether the release plan is appropriate and whether external assets are properly acknowledged.
If a paper uses personal data or human-subject data, reviewers should look for an appropriate statement that relevant ethical standards or institutional procedures were followed. Serious ethical concerns should be raised to the Area Chair rather than handled only through the review text. Reviewers should also value a meaningful discussion of limitations or possible negative societal impact, although the absence of such a discussion alone should not automatically determine the decision.
Reviewers should note that external links are not to be used to expand the content of the submission. If code, videos, images, or other media are intended for review, they should be included directly in the supplementary files in anonymized form. If a submission relies on external links in a way that may compromise anonymity or bypass review constraints, the issue should be reported to the Area Chair.
11. Timeline for Reviewers
- July 19, 2026 — Papers assigned to reviewers
Assigned submissions become available for review. Please check them promptly and notify the Area Chair as soon as possible if you identify a conflict of interest or a serious expertise mismatch. - By August 12, 2026 — Reviews due
Complete and submit your reviews by the review deadline. Reviews should be technically sound, sufficiently detailed, and constructive. - August 13-25, 2026 — Emergency review period
During this period, some papers may require follow-up, clarification, or additional reviews. Reviewers may be contacted if further input is needed. - August 26, 2026 — Reviews released to authors
Submitted reviews are released to the authors. - By September 2, 2026 — Author rebuttals due
Authors submit their responses to the reviews. Reviewers should be prepared to read the rebuttals carefully once they become available. - From September 2, 2026 — Discussion phase begins
Reviewer discussion starts after rebuttals are submitted. Reviewers are expected to participate in the discussion in a timely and constructive manner. - By September 8, 2026 — Final reviewer recommendations due
Reviewers should complete the discussion phase and submit their final recommendations by this date. - September 20, 2026 — Final decisions released to authors
Notification decisions are released to authors