Ends Dec 1, PT
SciFig
Conferences·

CVPR 2027 Poster Template and Guidelines

CVPR 2027 poster guide: the 2026 size (84 × 42 in, 2:1 landscape), official poster templates, PNG and video upload, LLM Usage Statement, and figure tips.

SciFig Team
SciFig Team
SciFig editorial team

CVPR 2027 runs June 20–25, 2027, in Seattle, Washington; the venue has not been announced yet.

This is a practical guide to the official poster and presentation rules for the IEEE/CVF Conference on Computer Vision and Pattern Recognition: poster size and template, the poster image and video upload, oral talks, and the figures a computer vision poster needs.

It also covers the CVPR 2027 LLM policy for authors and where accepted papers are published. Every date and specification below links to official CVPR materials; any prior-year planning reference is labeled with its year.

Preparing a talk as well as a poster? CVPR's own poster templates are linked in §2; for the slides, download our scientific presentation templates, editable 16:9 academic decks ordered the way a conference talk runs.
CVPR 2027 poster guidelines cover: a wide-angle view down a large computer vision poster hall with rows of posters on display boards and attendees with lanyards (Figure generated with SciFig)
CVPR 2027 poster guidelines cover: a wide-angle view down a large computer vision poster hall with rows of posters on display boards and attendees with lanyards (Figure generated with SciFig)

1. CVPR 2027 at a Glance: Dates, Location, and Paper Timeline

CVPR 2027 takes place June 20–25, 2027, in Seattle, Washington: workshops and tutorials run June 20–21 and the main conference sessions run June 22–25, with final paper decisions due February 25, 2027. The CVPR 2027 home page describes the meeting as the premier annual computer vision event, made up of the main conference and several co-located workshops and short courses. It names Seattle but does not name a venue yet.
The timeline below comes from the CVPR 2027 dates page and the CVPR 2027 Call for Papers. Paper deadlines are Anywhere on Earth (AOE), and the Call for Papers marks the enrollment and submission dates as fixed, with no extensions.
DateCVPR 2027 milestone
November 10, 2026Paper enrollment deadline
November 16, 2026Paper submission deadline
November 23, 2026Supplementary materials deadline
January 25 – February 1, 2027Reviews released; rebuttal period
February 25, 2027Final decisions
June 20–21, 2027Workshops and tutorials
June 22–25, 2027Main conference sessions
Submission goes through the CVPR 2027 OpenReview site, and every co-author needs an up-to-date OpenReview profile. CVPR has not posted a 2027 poster printing or upload deadline yet. In 2026, the official printing service took online orders until May 29, 2026, and the website upload asked for videos by May 20 and poster images by June 3. Treat those 2026 dates as a planning reference and check the CVPR 2027 dates page for the 2027 schedule.

2. CVPR 2027 Poster Size and Template: The 2026 Rules to Plan With

In 2026, CVPR posters were 84 × 42 in (213 × 107 cm), landscape, with a 2:1 aspect ratio, and workshop posters were limited to half that size, 42 × 21 in. CVPR has not published its 2027 poster instructions yet, so use these 2026 rules as a planning reference rather than as the 2027 CVPR poster size.
The CVPR 2026 poster printing information set out these poster guidelines:
CVPR poster question2026 rule (planning reference for 2027)
Main conference poster size (including Findings)84 × 42 in = 213 × 107 cm, width × height, 2:1 landscape
Workshop poster sizeHalf size, 42 × 21 in; main-conference authors may also use this size
Recommended layoutThree or four columns, little text, a few large and expressive figures
PurposeNot a copy of the paper; the tools for a 5–10 minute presentation to any attendee
File for the official printing servicePDF preferred; 100 DPI or vector art at full size; no bleed
Official printing servicePrinted on satin poster paper and delivered rolled to the convention center; the uploaded file is final

Those official 2026 instructions call for few, large figures rather than dense text, so the poster usually gets built around one method figure and one or two results panels. A 2:1 board is much wider than a standard 4:3 or A-series sheet, which leaves room for that dominant figure in the middle with the problem statement on the left and results on the right.

Build the hero figure for your CVPR poster

Start with the method figure that carries your paper, then lay out the wide 2026-size board (84 × 42 in) around it.

Explore the Tool
CVPR does provide official templates. The 2026 printing page linked "logos and poster templates" for main-conference, Findings, and workshop posters; download the current files from the CVPR poster printing page rather than from a copy found elsewhere, and look for the 2027 version once CVPR posts it. If you start from your own layout instead, resize it to 84 × 42 in (or 42 × 21 in for a workshop) before you export. For your other talks and posters, SciFig also has editable conference poster templates and scientific presentation templates, including lab meeting and journal club decks.

3. CVPR 2027 Poster Upload: PNG Size, Thumbnail, and 5-Minute Video

In 2026, each main-conference poster had a web page on cvpr.thecvf.com, and presenters were asked to add a 5-minute YouTube video, a poster thumbnail, and optionally a full poster image as a PNG of at most 5120 × 2880 pixels and 10 MB. The rules come from the CVPR 2026 video and poster upload instructions, which applied to the main conference only.

The 2026 upload instructions add these practical points:

  • Video first. Presenters linked a 5-minute YouTube video to their poster page by May 20, 2026 AOE, even if the poster and video were not final; both could be updated at any time. Where YouTube is unavailable, no video was required.
  • Thumbnail. An eye-catching graphic from the paper, shown on the poster page.
  • Full poster image. Optional but helpful for virtual attendees; upload by June 3, with re-uploads allowed afterwards.
  • Orals. Papers accepted as orals had both a poster page and an oral page; the video linked on the poster page appeared on the oral page automatically.
  • Findings and workshops. Findings papers were poster only and did not need uploads, and video and poster uploads were not supported for workshop papers.
A 2:1 poster is wider than the 16:9 upload frame, so a full-width export at 5120 × 2560 pixels keeps the printed proportions without padding. Export the PNG from the same file you send to the printer. CVPR has not published its 2027 upload rules yet; check the CVPR 2027 site for the 2027 version.

4. CVPR 2027 Oral Presentations and Official CVPR Logos

CVPR's published 2026 pages give no oral talk length, slide aspect ratio, or official slide template; what they do set is that at least one author of every accepted paper registers and presents the paper at the conference. The CVPR 2027 Author Guidelines repeat that attendance rule for 2027. CVPR has not published 2027 instructions for oral presenters yet; check the CVPR 2027 site for the CVPR 2027 presentation format.
For branding, the official logos sit alongside the poster templates on the CVPR poster printing page. Use those files rather than a logo copied from another site, so your poster or title slide carries the current version.
An oral talk is usually built from the same figures as the poster, so it helps to start both from one source file. Our scientific presentation templates give you 16:9 academic decks with the title, method, results, and conclusion slides already in order.

5. What Figures Do You Need for a CVPR 2027 Poster or Talk?

In our analysis, most CVPR posters rest on five core figure types: a teaser or concept figure, a model architecture diagram, a data and training workflow, qualitative result grids from your own model outputs, and quantitative plots or tables. The same figures usually have to work in the eight-page paper, on the poster board, and on slides. The CVPR 2027 Author Guidelines limit papers to eight pages including figures and tables, in the CVPR style from the CVPR 2027 Submission Template, with extra pages allowed only for references and the LLM Usage Statement.

Here is how each CVPR poster figure type maps to a figure workspace:

  • Model architecture diagram: the backbone, heads, losses, and where features are fused or upsampled. Start one in the neural network diagram generator, which builds it in SciFig FlowChart (flowcharts) with every block and connection editable.
  • Teaser or concept figure: the problem setting or the idea behind the method, such as a camera ray through a scene or a before-and-after task sketch. SciFig AI Illustration (AI scientific illustration) can generate it from a text description or your paper PDF.
  • Data and training workflow: data collection, annotation, training, and evaluation steps. SciFig FlowChart turns a written description into native SVG boxes and arrows.
  • Quantitative results: accuracy or mAP curves, ablations, and benchmark comparisons. These come from your measured numbers, so SciFig DataChart (charts and graphs from your data) plots them from your table.
  • Qualitative results: image grids, segmentation masks, and renderings come from your own model outputs on real inputs; use those actual outputs rather than generated pictures.
CVPR 2027 poster figure for a segmentation paper: a U-Net architecture diagram with an encoder contracting path, a bottleneck, a decoder expanding path, copy-and-concatenate skip connections, and a segmentation map output (Figure generated with SciFig)
CVPR 2027 poster figure for a segmentation paper: a U-Net architecture diagram with an encoder contracting path, a bottleneck, a decoder expanding path, copy-and-concatenate skip connections, and a segmentation map output (Figure generated with SciFig)
The U-Net diagram above shows the level of detail a CVPR audience reads closely: each skip connection joins an encoder stage to the decoder stage at the same resolution, and a legend names every arrow type once. For the conventions readers recognize in network figures, see our guide to drawing neural network diagrams.

Draw your network as an editable architecture diagram

Describe the backbone, heads, and connections once, then adjust any block or arrow until it matches your implementation.

Try Free

6. Designing a Winning CVPR 2027 Poster: Figures a Vision Audience Will Check

In our view, a winning CVPR poster lets one method figure carry the story at a glance, and every input, output, and arrow in it holds up when someone who works on the same problem traces it. CVPR's own 2026 guidance points the same way: few large figures, little text, and enough on the board to support a 5–10 minute conversation.
CVPR 2027 poster figure for a 3D vision method: a neural radiance field schematic sampling points along a camera ray, an MLP predicting color and density, and volume rendering into a pixel compared with ground truth (Figure generated with SciFig)
CVPR 2027 poster figure for a 3D vision method: a neural radiance field schematic sampling points along a camera ray, an MLP predicting color and density, and volume rendering into a pixel compared with ground truth (Figure generated with SciFig)

Three details are worth checking on any CVPR poster figure:

  • Inputs go where the method uses them. In a neural radiance field figure, viewing direction feeds only the color branch, while density depends on position alone. An arrow from viewing direction into density describes a different model.
  • Rays and rendering order. Points are sampled along a ray that starts at the camera and passes through one pixel, and volume rendering composites them from the camera side outward into that pixel's color.
  • Skip connections match resolution. In an encoder–decoder network, each skip joins two stages at the same spatial size; a skip drawn between mismatched levels describes a different architecture.
A general-purpose image model such as Nano Banana Pro or GPT Image returns one finished picture: when an arrow points the wrong way, you regenerate and hope the next attempt fixes it without breaking something else. In SciFig, the figure opens in an editor where every arrow, block, and label can be selected and changed, so you can reroute an input or fix a skip connection in place. Color is the other common slip on a results panel: distinguish methods by marker or line style as well as hue, and use a color-blind-safe palette such as Okabe–Ito.
Before printing, run the poster through the five common scientific figure mistakes checklist. If you also present at ICLR, our ICLR poster guidelines cover that conference's board sizes and policies.

7. Create Your CVPR 2027 Figures with SciFig

SciFig covers the three figures a CVPR poster or talk usually needs — SciFig AI Illustration (AI scientific illustration) for the teaser and method concept, SciFig DataChart (charts and graphs from your data) for quantitative results, and SciFig FlowChart (flowcharts) for the model architecture — and every result stays editable.
What you need for your poster or slidesUseWhat it helps with
Explain your method, the problem setting, or a teaser figureSciFig AI IllustrationStart from text, PDF, sketch, photo, reference, or old figure; edit any element; export PPTX or SVG
Turn results into mAP curves, ablation bar charts, or benchmark comparisonsSciFig DataChartUpload CSV or Excel for a publication chart; change chart type and colors; export the data chart as PNG or PDF
Show the model architecture or data workflowSciFig FlowChartGenerate from text or a template; edit text, boxes, and connectors; export SVG or PPTX
CVPR 2027 SciFig workspace mockup for a computer vision poster: a chat panel generating an editable two-stage object detector figure on the canvas, mAP-curve DataChart and dataset-workflow FlowChart cards, and PPTX, SVG, PNG, and JPEG export buttons (Figure generated with SciFig)
CVPR 2027 SciFig workspace mockup for a computer vision poster: a chat panel generating an editable two-stage object detector figure on the canvas, mAP-curve DataChart and dataset-workflow FlowChart cards, and PPTX, SVG, PNG, and JPEG export buttons (Figure generated with SciFig)

Spelling out the topology tells SciFig exactly which structure to draw. Copy and adapt:

two-stage object detector: input image to backbone, backbone to feature map,
feature map to region proposals, feature map and region proposals to RoI
pooling, RoI pooling to a class head and a box head.

Start from your CVPR paper, a prompt, or a sketch

Pick the source you already have, then edit every block and connection and export to PPTX or SVG for the poster and slides.

Try Free

8. CVPR 2027 LLM Policy: The Mandatory LLM Usage Statement and Your Figures

CVPR 2027 requires every paper to include an "LLM Usage Statement" that discloses any use of LLMs or other generative AI tools, names the tools, and identifies the affected parts of the work, which can include figures; papers without it will be desk-rejected. The CVPR 2027 Author Guidelines place the statement after the main text and before the references, and it does not count toward the page limit.

The policy sets out what the statement must do and who is responsible:

  • Disclose and describe. Name the tool or tools used, identify the parts of the paper or research affected (the guidelines give text, figures, code, experiments, related work, or ideation as examples), and briefly describe the nature and extent of the use.
  • Say so if none were used. In that case the statement reads "No LLMs or generative AI tools were used in this work".
  • Authors stay responsible. Authors may use LLMs and other AI tools but are fully responsible for everything they submit, and existing standards on research integrity, attribution, reproducibility, and image integrity apply equally to human- and AI-generated content. An LLM should not be listed as an author.

The policy covers the paper; CVPR has not published a separate AI rule for posters or slides. If you make a figure with SciFig, you can name it in the LLM Usage Statement or in the figure caption. For example: Figure created by the authors using SciFig (scifig.ai). Use it only when it describes how you actually made the figure.

9. Where CVPR 2027 Papers Are Published: CVF Proceedings

CVPR's official pages list no abstract book or journal supplement; according to the CVPR 2027 Call for Papers, all accepted papers are publicly released by the Computer Vision Foundation (CVF) two weeks before the conference. The CVF hosts them on its open access proceedings site, and the CVPR site also links the IEEE proceedings, so a reader who finds your poster will usually open the CVF page next.

Two details about that record matter for poster authors:

  • Public disclosure date. The Call for Papers notes that a paper's official public disclosure date is two weeks before the conference, or the date the authors make the paper public themselves, whichever comes first.
  • Submissions stay confidential until then. Submissions, reviews, and rebuttals are not made public, and papers must not be discussed with the media mentioning CVPR until they are accepted.

Because the CVF page is what people cite, make sure the architecture figure on your poster matches the one in the camera-ready paper.

Frequently Asked Questions


References

All dates, specifications, and policies above are drawn from official CVPR materials, accessed November 15, 2026:
  1. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2027. cvpr.thecvf.com. Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2027
  2. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2027 Meeting Dates. Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2027/Dates
  3. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2027 Call for Papers. Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2027/CallForPapers
  4. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2027 Author Guidelines. Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2027/AuthorGuidelines
  5. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2026 (2026 conference site). Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2026
  6. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2026 Poster Printing Information (2026 rules). Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2026/PosterPrinting
  7. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2026 YouTube Video and Poster Uploads (2026 rules). Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2026/YouTubeVideoPosterUpload
  8. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2026 FAQ (2026 rules). Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2026/FAQ
  9. IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR 2026 Author Guidelines (2026 rules). Accessed November 15, 2026. https://cvpr.thecvf.com/Conferences/2026/AuthorGuidelines
  10. OpenReview. CVPR 2027 Conference. Accessed November 15, 2026. https://openreview.net/group?id=thecvf.com/CVPR/2027/Conference
  11. Computer Vision Foundation. CVF Open Access. Accessed November 15, 2026. https://openaccess.thecvf.com/menu

Disclaimer: This article is educational content about scientific figure design for CVPR posters and presentations. SciFig is an independent AI scientific figure platform covering illustrations (SciFig AI Illustration), charts and graphs from your data (SciFig DataChart), and flowcharts (SciFig FlowChart) and is not affiliated with or endorsed by CVPR, IEEE, or the Computer Vision Foundation.

Keep going

Recommended resources

Templates, tools, and further reading for your research

Call to action background

Ready to start?

Publication-ready scientific figures, in minutes

Start Creating Free

Free to start · No credit card required · Built for researchers