International Journal of Computer Vision Submission Guide

Source: IJCV author guidelines, aims and scope and the journal home page.

Use this page when deciding:

• whether the technical contribution is substantial enough for IJCV

• whether experimental validation meets IJCV's bar

• whether the conference-to-journal extension is sufficient

• a clear technical contribution beyond conference version

• comprehensive experimental validation against state-of-the-art

• mathematical or algorithmic novelty

• reproducibility materials

• a cover letter quantifying new contributions

• Insufficient extension beyond conference version.

• Missing comprehensive baseline comparisons.

• Engineering applications without theoretical contribution.

• Thin reproducibility materials.

IJCV is a flagship computer-vision journal.

Theory + experiment requirement: the journal differentiates from CVPR/ICCV/ECCV conference papers by demanding deeper analysis and comprehensive experiments.

Conference-extension expectation: IJCV expects journal versions to make a clear journal-level contribution beyond conference versions.

Early editor screen: the first pass is decisive because IJCV editors need to see a journal-level computer-vision contribution before external reviewers invest months in the paper.

The strongest IJCV cover letters establish:

• the technical contribution

• the substantial extension beyond conference version

• the experimental validation scope

• the theoretical novelty

International Journal of Computer Vision submissions go through Springer Editorial Manager at Editorial Manager submission portal. The guidelines recommend that authors use their ORCID ID when submitting an article. The journal publishes Regular, Short and Survey articles on computer vision research. Full guide at IJCV Submission Guidelines.

Springer Nature's IJCV guidelines ask for the following at or around submission:

• A cover letter to the Editor that mentions re-used material and unpublished material (such as manuscripts in press) and, as Manusights recommends, explicitly establishes the technical contribution and the relationship to prior conference-version work (if applicable)

• Declaration of competing interests for all authors

• Data availability statement covering benchmark datasets used and any new datasets released

• A code availability statement where code supports the results (Springer Nature lists material and/or code availability among its declarations)

• Authors' contribution statement (Springer Nature shows CRediT as one example format)

• Optionally, suggested reviewers (and any requested exclusions): they must be totally independent and not connected to the work, with an institutional email address for each

• For conference-version extensions (a Manusights recommendation, not a stated IJCV requirement): explicit difference statement documenting substantial new content beyond the conference paper and line-level differences from CVPR, ICCV, ECCV, or workshop versions

For International Journal of Computer Vision submissions, the most common artifact-related issue is insufficient new content for conference-extension submissions. Submissions that mostly recapitulate a CVPR, ICCV, or ECCV paper without substantial new experiments, theory, broader benchmarks, or reproducibility support are commonly returned before the author gets a useful reviewer conversation.

For International Journal of Computer Vision submissions, the editorial timeline runs through four phases. The journal's home page lists a median of 37 days from submission to first decision, computed over all submitted manuscripts including those not put into the review pipeline at the onset, and says the typical time to first decision for manuscripts is approximately 96 days. The journal says reviewers are given a tight deadline and that timing can fluctuate with article length, reviewer availability when the topic is narrow, and the depth of reviewing needed.

Editorial Manager intake and editor assignment

Submission goes through Editorial Manager, which compiles the uploaded source files into a single PDF for the author to approve. The journal says a complete set of editable source files is needed or the article will not be considered for review, and that submissions without the relevant declarations are returned as incomplete. The journal does not publish an editor-assignment time or a conference-extension check; in our experience the most common early hold-up is insufficient documentation of new content beyond a prior conference paper.

Editor scope and significance screen

IJCV's editor filter prioritizes technical contribution plus computer-vision significance. A common weak pattern in our editorial research is a paper that reads like an incremental refinement of a benchmark result without a clear conceptual or methodological advance. The official source set reviewed here did not publish stable acceptance-rate or desk-rejection-rate figures, so this guide does not quote them.

Peer review

The journal publishes no reviewer count and no peer-review duration; it says reviewers are given a tight deadline and that timing depends on reviewer availability when the topic is narrow. Reviewer mix typically includes one methods expert plus one application-domain or evaluation specialist. Submissions missing comparison against current state-of-the-art benchmarks or ablation studies tend to extend reviewer dialogue.

Because the review window can leave the status bar unchanged for long stretches, it helps to know how to interpret IJCV's under-review status at each phase before assuming silence means trouble.

Decision and revision

The journal publishes no typical first-decision outcome, number of revision rounds or time to acceptance. Once an article is accepted, it says the article should be published online within 3-5 weeks if proofs are returned quickly.

• Is IJCV a good journal?

• IEEE TPAMI submission guide

Before upload, run your manuscript through an IJCV technical contribution readiness check.

Use the public instructions for mechanics, then look at the package the way an editor will. The review tells you whether your paper clears the International Journal of Computer Vision fit check before upload, especially around conference-extension statement that cannot prove a journal-level contribution, benchmark table that avoids the current state of the art, and algorithmic novelty asserted without being isolated. Paid Manusights reviews include a 60-day money-back guarantee, and we do not train models on submitted manuscripts.

Across computer-vision manuscripts targeting International Journal of Computer Vision, three manuscript-level patterns matter most.

Conference-extension statement that cannot prove a journal-level contribution

Across computer-vision manuscripts targeting International Journal of Computer Vision, the most common weak point is not the existence of a prior CVPR, ICCV, ECCV, NeurIPS, or workshop paper. It is the failure to prove what changed after that prior version. The abstract may still read like the conference paper, the contribution paragraph may list "additional experiments" without explaining why they alter the claim, and the cover letter may ask IJCV editors to infer the extension by comparing two PDFs.

For IJCV, the manuscript components need to do that work explicitly. The cover letter should map old contribution to new contribution. The introduction should name the new theoretical, algorithmic, empirical, or application-level advance. The methods and supplementary files should make the expanded experiment set auditable. The main tables should show the new evidence, not only repeat conference baselines.

If the difference statement reduces to longer proofs, more implementation details, or a few additional datasets, the paper may fit Computer Vision and Image Understanding, Pattern Recognition, IEEE Transactions on Image Processing, or a specialist application journal better than IJCV. A strong IJCV extension reads like a new journal article, not a conference appendix with a different template.

Check whether your IJCV extension case is visible enough for editors →

Benchmark table that avoids the current state of the art

Across computer-vision manuscripts targeting International Journal of Computer Vision, the second pattern appears in papers with a plausible algorithm but an underpowered evidence package. The abstract claims a state-of-the-art contribution, but the main benchmark table compares against older baselines, omits the strongest recent transformer, diffusion, 3D reconstruction, segmentation, tracking, or vision-language method, or hides unfavorable comparisons in supplementary material. IJCV reviewers know the benchmark landscape and will ask why the comparison set was chosen.

The fix is to make the evidence package harder to dismiss before upload. The methods should define datasets, splits, preprocessing, annotation assumptions, and computational budget. The figures should show failure cases, not only best examples. The tables should include current competitive baselines, ablations that isolate the claimed mechanism, and complexity or runtime where deployment relevance is part of the claim. The supplementary files should include code, trained-model details, data availability, hyperparameters, and seed sensitivity where possible.

If the manuscript cannot withstand that comparison set, the better move may be to retarget IEEE Transactions on Image Processing, Machine Vision and Applications, Pattern Recognition Letters, or Computer Vision and Image Understanding while the stronger IJCV evidence is built.

Check whether your IJCV tables and ablations can survive current-baseline review →

Algorithmic novelty asserted without being isolated

Across computer-vision manuscripts targeting International Journal of Computer Vision, the third pattern is a paper that reports a performance gain but does not isolate why the gain exists. The model may be larger, the training data may be cleaner, the augmentation stack may be stronger, or the evaluation may have shifted, but the abstract and methods still claim a new algorithmic contribution. That leaves IJCV editors with a trust problem: is the paper advancing computer vision, or only reporting an engineering recipe that happened to score higher?

The manuscript components that fix this are concrete. The abstract should state the mechanism of improvement, not only the result. The methods should separate architecture, objective, data, training schedule, and inference changes. The ablation table should show which component carries the effect. The figure sequence should make the algorithm interpretable enough for a reader outside the exact subarea. The discussion should acknowledge failure modes, compute limits, dataset bias, and transfer boundaries.

If the novelty is primarily engineering integration, IEEE Access, Applied Intelligence, Pattern Recognition, or an application-specific computer-vision venue may be the more credible route. IJCV needs the package to show why the method changes understanding, not only leaderboard placement.

Check whether your IJCV novelty claim is isolated in the abstract, methods, and figures →

The review tells you whether your paper clears the IJCV journal-level contribution check before you spend another cycle on upload mechanics. Manusights checks do not train on your manuscript, and paid reviews include a 60-day money-back guarantee when the report does not meet the stated review scope.

Of the International Journal of Computer Vision manuscripts we pre-screen, the recurring pattern is a conference paper extended in length rather than in depth: IJCV is the journal venue, so editors and reviewers expect a contribution whose theory, generalization, or analysis genuinely needs the longer format, not a CVPR or ICCV paper padded with extra experiments. Submit if your work offers a formulation or study whose depth justifies a journal article; think twice if the extension is supplementary material promoted into the main body.

This page was researched from IJCV's author guidelines, Springer editorial-policy materials, Clarivate JCR data, and Manusights editorial research for computer-vision submissions. Across the Manusights editorial review for this page, the recurring fit issue was whether the abstract, methods, benchmark tables, ablations, supplementary material, and cover letter prove a journal-level computer-vision contribution rather than only a conference-paper extension.

In IJCV-focused pre-submission diagnostics, we consistently see four signals that distinguish strong submissions from weak ones. First, the journal version must add substantial new content beyond any prior conference paper. Second, experimental validation should cover state-of-the-art baselines on standard benchmarks. Third, theoretical contribution should be clearly stated. Fourth, reproducibility materials should be available and described in the methods or supplementary files.

For International Journal Of Computer Vision-targeted manuscripts, the single most consistent feedback class we deliver in pre-submission diagnostics for IJCV is the conference-extension distinction. IJCV expects journal versions to make a journal-level contribution beyond conference versions. Submissions that primarily reformat conference papers routinely receive "insufficient extension" feedback during desk screening. We coach authors to articulate the new contributions explicitly. If the new contributions reduce to "we provide more details," the extension is structurally weak.

If they read like "we add a new theoretical analysis showing X, prove convergence under Y assumptions, and demonstrate generalization to Z domain," the extension is structurally substantial.

For International Journal Of Computer Vision-targeted manuscripts, beyond the rubric checks, three pre-submission diagnostic patterns recur most often in the manuscripts we review for IJCV. First, manuscripts where the contribution section uses generic language without specifying baseline comparisons are flagged at desk for insufficient detail. Second, manuscripts that lack engagement with the journal's recent issues are at risk of being told the contribution doesn't fit the publication conversation. Third, manuscripts with reproducibility materials marked as "available upon request" are increasingly flagged.

For IJCV-targeted manuscripts, the strongest packages usually do three distinctive things before upload: they state the journal-level contribution in the first cover-letter paragraph, they include a precise difference map from any CVPR, ICCV, ECCV, NeurIPS, or workshop version, and they point to recent IJCV articles that define the technical conversation the new manuscript is joining.

Beyond the rubric checks, IJCV editorial triage operates on limited time per manuscript. Editors typically scan the abstract, introduction, contributions section, and experimental tables before deciding whether to invite reviewer engagement. Manuscripts that bury the technical contribution or require multiple readings to identify the central advance fare worse than manuscripts that lead with their strongest signal.

We coach researchers to design the abstract, contributions section, and experimental tables for fast assessment: each independently conveys the contribution, the methodological rigor, and the empirical performance, rather than relying on linear reading of the full manuscript.