# CVPR 2026 review form > Simulate CVPR 2026-style reviews: Summary, Strengths, Weaknesses, a five-level Preliminary Recommendation (Strong Accept to Reject) and a justification with points for the rebuttal, written in CVPR's depersonalized voice. - Venue: IEEE/CVF Conference on Computer Vision and Pattern Recognition - Form status: modeled, not official. The CVPR review form is not public. Field order is reconstructed from the 2024–2025 OpenReview form; the five recommendation labels come from the 2026 reviewer training material, and the 1–5 numeric mapping is ours. - Official source: https://cvpr.thecvf.com/Conferences/2026/ReviewerTrainingMaterial - Overall score: Preliminary Recommendation, 1–5 - Rough acceptance bar (our simulation): mean ≥ 3.5 - Web page: https://reviewpaper.ai/venues/cvpr ## How reviewers read the paper CVPR's training material asks for a summary in the reviewer's own words, specific evidence-based strengths and weaknesses (novelty concerns must cite the specific prior work), and a recommendation with a justification that tells the authors what the one-page rebuttal should address. Do not reject solely for missing SOTA, missing comparisons to arXiv-only work, easily fixable errors, or a missing limitations section. Refer to "the paper", not "you" or "the authors". - Refer to "the paper" — never "you" or "the authors". - Professional, courteous tone with concrete suggestions. - Short or generic reviews are flagged as irresponsible; each weakness needs specifics. - Do not request new experiments as a condition for the rebuttal (the rebuttal cannot add new experiments or contributions). **Where the bar sits.** CVPR 2026 accepted about 25% of 16,092 submissions. No score statistics are public; on this 1–5 mapping, a mean of ≥ 3.5 (mostly Weak Accept or better) is a rough proxy for acceptance. ## Reviewer form, field by field ### 1. Summary Type: free text. 2–4 sentences in your own words: what the paper does and claims. ### 2. Strengths Type: list of points. Specific and evidence-based. Avoid generic praise such as "the paper is interesting". ### 3. Weaknesses Type: list of points. Specific and detailed — bullet lists of short phrases are not a detailed review. A novelty concern must cite the specific prior work and explain how closely it relates. Do not reject solely for not beating SOTA, for lacking comparisons to arXiv-only work, or for easily fixable errors. ### 4. Preliminary Recommendation Type: score, 1–5. Strong Accept: significant contribution with strong evidence and broad interest. Weak Accept: sound and well motivated but incremental or not best on every benchmark. Borderline: reasonable but with gaps in evaluation or positioning. Weak Reject: limited novelty relative to closely related prior work. Reject: fundamental technical issues. - **5**: Strong Accept - **4**: Weak Accept - **3**: Borderline - **2**: Weak Reject - **1**: Reject ### 5. Justification For Recommendation And Suggestions For Rebuttal Type: free text. Connect the rating to the assessment above, and state precisely what the rebuttal should address. ### 6. Confidence Level Type: score, 1–5. How certain you are about your assessment. - **5**: You are absolutely certain about your assessment. You are very familiar with the related work and checked the math/other details carefully. - **4**: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. - **3**: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked. - **2**: You are willing to defend your assessment, but it is quite likely that you did not understand the central parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked. - **1**: Your assessment is an educated guess. The submission is not in your area or the submission was difficult to understand. Math/other details were not carefully checked. ### 7. Dataset Attribution Type: free text. Are all data assets properly cited? Flag any use of withdrawn or deprecated datasets. "No concerns" otherwise. ### 8. Ethics Flag Type: free text. "No" or a brief description of the ethical concern (e.g. human-subjects data without IRB statement). ## Area chair meta-review CVPR ACs work in triplets; the meta-review must justify the recommendation from the reviews and discussion. Recommendation options: Award candidate · Oral · Highlight · Poster · Findings · Reject ## CVPR author requirements 1. Main text within 8 pages plus references 2. Anonymized submission 3. Negative societal impact discussed where relevant 4. IRB approval stated for human-subjects data 5. All datasets cited; no withdrawn datasets used 6. No hallucinated or unverifiable citations ## LLM policy CVPR 2026 does not allow LLMs to be used to write reviews or meta-reviews, and no substantial paper or review content may be shared with an LLM. Violations can lead to desk rejection of the reviewer's own papers. Official policy: https://cvpr.thecvf.com/Conferences/2026/ReviewerGuidelines --- Simulate reviews of your own draft in this venue's form: https://reviewpaper.ai/new?venue=cvpr (3 free reviews). Sample report: https://reviewpaper.ai/sample. reviewpaper.ai is for authors checking their own unpublished drafts. If you are an assigned reviewer, follow your venue's LLM policy: most venues forbid putting a submission you are reviewing into any external AI tool. Not affiliated with any conference.