International Journal of Computer Vision Submission Guide

Source: Clarivate JCR 2024, Springer editorial disclosures (accessed April 2026).

Source: IJCV author guidelines.

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. Initial setup requires a Springer Nature account; ORCID is recommended for the corresponding author. The platform accepts unsolicited Regular Papers, Short Papers, and Survey articles on computer vision research. Full guide at IJCV Submission Guidelines.

International Journal of Computer Vision requires these at first submission:

• Cover letter explicitly establishing 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

• Code availability statement with public repository link for any methods that should be reproducible

• CRediT author contributions statement

• Four or more suggested reviewers with no recent collaboration history

• For conference-version extensions: 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 over an unusually long total window. SciRev community data and documented author reports indicate median first decisions at 4-6 months, reflecting the journal's rigorous reviewer-pool selection and the time needed to evaluate substantial computer-vision contributions.

Day 0 to 14: Editorial Manager intake and editor assignment

Springer Nature intake handles format compliance plus the conference-extension new-content check. The handling Editor assignment lands within 14 days; vision papers route to subject editors matching the technical subfield (deep learning architectures, 3D vision, vision-language models, video understanding, low-level vision). The most common Day 0-14 hold-up: insufficient documentation of new content beyond a prior conference paper.

Week 2 to 6: Editor scope and significance screen

IJCV's editor filter prioritizes technical contribution plus computer-vision significance. The common Week 2 to 6 weak pattern in 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.

Week 8 to 24: Peer review

At least 2 reviewers per IJCV policy, often 3-4 for substantive papers. The 4-6 month first decision target reflects difficulty in securing qualified reviewers in active CV subfields. 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 extend reviewer dialogue by 6-10 weeks.

Week 24 to 52: Decision and revision

Major revision is the standard first decision at IJCV. Revision rounds typically settle at 2, occasionally 3 for accepted papers. Total submission-to-acceptance: 8-14 months for accepted papers (long compared to CV conference cycles but reflective of the journal's review depth).

• Is IJCV a good journal?

• IEEE TPAMI submission guide

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

Read the public instructions for mechanics, then pressure-test the package the way an editor will see it. 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.

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.