Guide
WACV vs BMVC: round 1 is not a free shot
Authors treat WACV's first round as a rehearsal they can retake in the second, and WACV desk-rejects that move. What round 1 buys is a revision the same reviewers judge, which nothing else in the CVF calendar sells.
Published 14 August 2026, dates rechecked 10 September 2026
Waiting on 9 October, or past both rounds? Both WACV 2027 rounds are shut and both are answered on 9 October, so the other rooms are worth ranking now. Jump to the matcher
The 1,448 and the 404 are BMVC's own, on its accepted papers page; the percentage is our arithmetic, because BMVC states none. The 831 is our count of the entries on WACV's 2026 accepted papers listing on 14 August 2026. We found no submission total for any WACV edition, so no rate follows.
The short version
- WACV runs two rounds and they are not two attempts. A round-1 reject sent back as a new round-2 paper is desk rejected, in WACV's own words.
- A round-1 Revise and Resubmit is the one that carries over. It keeps its ID, takes a one-page rebuttal and a revised PDF, and goes back to the same reviewers and chairs. Round 2 has neither.
- So send rough but real work in round 1 and finished work in round 2. For WACV 2027 both rounds are shut, and 9 October 2026 returns round-1 final decisions and round-2 decisions together.
- BMVC, the other half of the question, is shut. 404 of 1,448 accepted, notified 7 August 2026, and no 2027 call yet.
The mechanism
Round 1 sells a revision. Round 2 sells one shot.
WACV names the benefit itself: the primary benefit of submitting in Round 1 is that submissions can be invited for resubmission to Round 2. Misreading it costs a desk rejection.
| Round 1 | Round 2 | |
|---|---|---|
| Enrolment and reviewers | At least three reviewers, consolidated by a pair of chairs. | The same panel process, and round-1 papers are enrolled for it automatically. |
| Accept | In. Camera ready 2 November 2026. | In, same camera-ready date. |
| Revise and Resubmit | "The submission has not been accepted to WACV 2027, but may be made acceptable after some revisions." The paper keeps its ID; do not create a new submission. | No such outcome exists. |
| What you may send back | A one-page rebuttal, a revised paper highlighting what changed, and optional revised supplementary. One deadline, which fell on 28 August 2026. | Nothing. "Round 2 papers will not have a rebuttal and revision step." |
| Who judges the fix | "The revised version will be sent to the same reviewers and ACs, after which the final decisions will be made." | Not applicable. The chairs decide on the reviews directly. |
| Reject | Made when the reviews raise substantial revisions that cannot be addressed in a short time frame. | Final. Decisions 9 October 2026 either way. |
| A round-1 reject entering round 2 | "Rejected round 1 papers which are submitted to round 2 as new papers will be desk rejected." | So the retake most authors are planning does not exist. |
Quoted from the WACV 2027 author guides, the pair meeting from the area chair guides, read 14 August 2026. Every deadline here is Anywhere on Earth.
So submit rough in round 1 and finished in round 2
Two rounds read like two chances, so the habit is to fire the first off early and keep the second in reserve. There is no reserve. Round 1 holds a middle verdict nothing else in this family offers, and who reads the fix is the difference: a CVPR, ICCV or ECCV rebuttal argues at reviewers who never see a changed paper, while a WACV revision is a new PDF read by the people who wrote the objection.
The two calendars
Both are shut, and one date WACV disagrees with itself about
WACV 2027 closed round 2 on 28 August 2026 and answers both rounds on 9 October. BMVC 2026 notified on 7 August, so you cannot submit to it at all.
| Stage | WACV 2027, round 1 | WACV 2027, round 2 | BMVC 2026 |
|---|---|---|---|
| Registration or abstract | 19 June 2026 | 21 August 2026, new papers only | 22 May 2026 |
| Paper, then supplementary | 26 June, then 28 June 2026 | 28 August, then 30 August 2026 | 29 May 2026 for both, and no extensions were granted |
| Reviews reach the authors | 7 to 9 August 2026, and the disagreement is WACV's own | Not before the decision | By 3 July 2026, when the rebuttal opened |
| The author's turn | Rebuttal and revision, one deadline, 28 August 2026 | None | 3 to 10 July 2026, reinstated after BMVC 2025 had none |
| Decisions | 9 October 2026 | 9 October 2026 | 7 August 2026 |
| Conference | 4 to 8 January 2027, at the Disney Springs | The same event | 23 to 26 November 2026, Lancaster Town Hall, in person |
| Can you still act on it | Closed. Nothing left to send | Closed on 28 August 2026. Nothing left to send either | Closed. Still no 2027 call as of 10 September 2026 |
WACV's deadline table gives 9 August 2026 for round-1 reviews and decisions to authors, against 7 August on the call for papers, the reviewer guidelines and both chair guides: four pages against one, so we print the range. BMVC dates from its call for papers and reviewing guidelines. Read 14 August 2026.
A date a venue prints two ways is a date to check yourself
Five WACV 2027 pages carry the round-1 decision date. Four say 7 August 2026, the deadline table says 9 August, and nobody has reconciled them: two days out of the three weeks round-1 authors had to rebut and rewrite. Treat any page quoting one confident WACV date as unchecked.
The rules side by side
Same four-page trap, half the pages, twice the reviewing
Where these two texts agree they agree almost word for word. Where they diverge, what you hand in changes.
| Rule | WACV 2027 | BMVC 2026 |
|---|---|---|
| Main paper | Eight pages including figures and tables, plus pages carrying only references. | 14 pages excluding references, raised for 2026, which is why the review period grew. |
| Supplementary | Videos, proofs, figures or code, but not results on additional datasets and not results from an improved version of the method. 200MB, and reviewers are not obligated to look. | Unlimited, one zip, filed with the paper. The same two bans. Code optional. |
| Rebuttal | One page, anonymous, and it cannot include links, even anonymous ones. Round 1 only. | A week, and reviewers are told not to ask for substantial extra experiments. No page limit is published. |
| What you owe as a reviewer | "All authors are obligated to serve as reviewers." Expect 3 to 4 papers per round and 6 to 7 in total, and late or thin reviews may desk-reject your own submissions. | Softer: a commitment to review a proportional number, up to 3 each. Chairs rate every review, and reviewers under the bar are not acknowledged as committee members. |
| What counts as a prior publication | arXiv preprints and technical reports do not. A peer-reviewed workshop paper longer than four pages excluding references does, and 20 percent overlap makes two submissions the same paper. | The same sentence, the same four-page threshold, the same 20 percent. |
| LLMs | Disclosure, after NeurIPS 2025: a paper must fully describe its methodology and any tool important to it, LLMs included. Citations of non-existent material will be rejected when found. | Liability, not disclosure: attributing plagiarism or inaccuracy to LLMs is not a viable defence. Reviewers may not use them beyond background research or checking short phrases. |
| The numeric review scale | Not published on any of the six WACV 2027 pages we read. | Not published on the author, reviewing or chair pages either. |
From the WACV author and reviewer guidelines and the BMVC author and reviewing guidelines, read 14 August 2026. Each rule binds the edition beside it.
Neither publishes a review scale, so we print none
We looked on six WACV 2027 pages and three BMVC 2026 ones for a rating scale, a confidence scale or the fields of the form, and found none at either. Both run on OpenReview, where the form sits behind a committee login, which is why CVPR, ICCV and ECCV publish none either. Any table of WACV scores against BMVC scores was reconstructed from a source its author cannot show you.
The tracks
WACV tells its chairs to overrule reviewers who demand algorithmic novelty
The sharpest thing either venue publishes, and it is aimed at exactly the paper that keeps getting turned down elsewhere: a strong system in a real domain with no new loss function.
| Track | Judged on | Explicitly not judged on |
|---|---|---|
| Applications | "systems-level innovation, novelty of the domain and comparative assessment", which means a new task, a new formulation of an existing one, or a new way of benchmarking existing methods. | "They should not be evaluated solely on the basis of algorithmic novelty (i.e, it is okay to have algorithmic novelty, but it is also okay not to have it)." |
| Algorithms | "the standard conference criteria including algorithmic novelty and quantified evaluation against current, alternative approaches", in the style of CVPR, ICCV and ECCV. | Not solely on systems-level innovation. The protection runs both ways, in the same wording. |
| Evaluations and Datasets | New for 2027, modelled on the NeurIPS track: tools, datasets, benchmarks and practices for testing and auditing AI and ML systems. | Reviewers are asked whether the dataset will meaningfully advance any particular sub-areas in Computer Vision, and whether the benchmarking is correct. |
| What the chairs are told | "In the past, some reviewers have recommended rejecting application papers based on their lacking algorithmic novelty." Chairs are asked to watch those reviews and "when needed, please overrule the reviewers' recommendations". | So ticking the wrong box hands your paper the wrong rubric and forfeits the protection written for it. |
Criteria from the WACV 2027 reviewer guidelines, the tracks from the call for papers, the escalation clause from the area chair guides, and BMVC's brave-concepts wording from its reviewing guidelines. Read 14 August 2026.
Pick the track before you write the introduction
The box you tick decides which of two published rubrics your paper is read under. An applications paper whose introduction sells a novel module invites the algorithms rubric, and with it the objection the applications rubric exists to stop. BMVC has nothing equivalent: it asks reviewers to embrace brave concepts untested on many datasets, but that is a tone instruction, not a track with a chair told to enforce it.
Free tool, no account
Where should this paper go?
Paste the title and the abstract. It ranks the venues that publish this kind of work, with the reach and the article charge OpenAlex has on record.
WACV answers both rounds on 9 October 2026
One date carries both: final decisions on round 1, and reviews plus a decision on round 2, with no rebuttal or revision left in either. If the answer is no, or if both rounds shut before the paper was ready, the question is which room takes the work instead. The matcher ranks every venue we hold against your abstract, free and with no account.
Coming out of the bigger vision venues instead: after a CVPR rejection. What ends a submission without a review, at whichever venue is next, is on the pre-submission checklist. The limits of a machine read are in what AI peer review cannot do.
phdflow reads a finished draft against the guidelines of the venue you are aiming at and returns an accept or reject call with the objections behind it. Scored against 697 real decisions: AUC 0.91 and 96% accuracy on 297 ICLR 2025 papers, one false accept in 149 rejects. The full measurement. Read a finished review before you pay for one. Your first paper gets a free preview; the full report is 60 credits, €9. No account needed. Pricing.