Computer Vision Conferences 2027: CVPR, ICCV, WACV

Computer vision has the clearest calendar in AI research, once you know the rhythm. One flagship every year, two that alternate, and a handful of smaller venues that are often better value than the big names. This guide pulls the dates from our AI events calendar rather than typing them, so what you see is current, and then explains which conference suits which kind of trip.

Every computer vision conference still to come in the calendar, in date order. Each name opens a page with the venue, the format and the organiser's site. Always confirm dates and deadlines there before you book anything.

ECCV is missing from the list because it runs in even years and its next edition after the current one is not in our data yet. BMVC and ACCV, the British and Asian conferences, are smaller and well regarded, and they will appear here when their dates are confirmed.

The pattern is simple and it shapes everyone's year, from PhD students timing their submissions to companies deciding where to put a recruiting stand.

CVPR runs every year, usually in June, and is the largest event in the field. It is co-sponsored by the IEEE and the Computer Vision Foundation. The days before the main conference are workshops and tutorials, and many people find those more useful than the main programme, because they are narrower and the rooms are smaller.

ICCV runs in odd years and is the other IEEE and CVF flagship. It moves around the world, so its location changes the travel maths for your team from one edition to the next. 2027 is an ICCV year.

ECCV runs in even years and is the European conference, though it is international in practice. Together, ICCV and ECCV mean there are two flagships every year: CVPR plus whichever of the other two is due.

WACV runs every winter, early in the year, and its full name tells you what it is for: the Winter Conference on Applications of Computer Vision. It takes applied and systems work that may not suit the flagships, and it is a fraction of their size.

The deadline matters more than the date. For anyone submitting, the conference is the end of a process that starts with the call for papers, many months earlier. The useful calendar for a research team is the submission calendar, so put each flagship's deadline, rebuttal period and camera-ready date in the team's planning as soon as the call is published. For everyone else, the conference date is what matters, and that is what the list above shows.

Vision work also turns up at the general machine learning conferences, NeurIPS, ICML and ICLR, and at the robotics ones, ICRA and IROS, where perception is half the programme. If your team works on embodied systems, the robotics and embodied AI conference guide is the better starting point, and the NLP and machine learning guide covers the general venues.

You are a researcher publishing new methods. CVPR and the year's ICCV or ECCV are the targets. Your choice is really about deadlines: which submission window matches when your work will be ready.

You are an engineer shipping vision products. WACV and the CVPR workshops. The papers are closer to the problems you have, the authors have time to talk, and you will meet other people deploying models in messy conditions rather than on benchmarks.

You lead a team and need to hire. CVPR, because that is where the volume of candidates is. A stand, a dinner, or just a well-organised list of meetings with authors whose papers match your problems. For most companies without a paper, recruiting is the business reason to attend a flagship.

You sell to vision teams. CVPR's expo, and possibly ICCV's. Your buyers are there and they are in a receptive mood. Be honest with yourself about the cost of a stand against the deals you could plausibly trace back to it.

You invest in or partner with vision companies.CVPR again, but spend your time at the workshops and posters rather than the expo. The founders of next year's companies are more often presenting a poster than standing behind a booth. Read the workshop programme for your area, find the papers with industry co-authors, and book coffee with the authors.

You run a business and want to understand AI. None of these. A research conference is the wrong room for that question. The guide to AI conferences for marketing and sales leaders and the US AI conference guide list events built for business readers.

The same conference is three different events depending on why you are there, and the mistake is to plan it as all three.

Teams that send several people often do best by splitting the jobs explicitly: one person owns the research list, one owns recruiting meetings, one owns the expo. Without that split, everyone ends up in the same keynote and nobody covers the workshops.

Research conferences are easier to justify than they look, as long as you state the business reason. A paper being presented is the strongest case, because the company's name is on the programme. Without one, the case rests on what you will bring back.

Name the outcome. "Meet eight candidates for the perception role" or "evaluate three labelling vendors" is a case. "Keep up with the field" is a request for a holiday.

Price the whole trip. Registration, flights, hotel, and the days away. Flagships are often long, and the workshop days add to it, so pick the days you actually need rather than the whole week.

Promise a write-up. A short internal note on the papers that matter to your work, the people you met and what you recommend, delivered within a week. That note is what makes the next request easy.

A month before: check visas if you are travelling internationally, because flagship venues move continent and slow visa processing is a familiar way to miss a conference you were accepted to. Book meetings with the people you want to see.

During: after each useful conversation, write one line about it. Not the name, which you will have. The specific thing: what they are working on, what they are looking for, what they offered to send you.

Within three days after: follow up. Candidates, prospects and collaborators all forget which of the hundreds of people they met you were, faster than you would think. The guide to event lead follow-up explains why the first 72 hours decide most of it. For companies that exhibit, this is where Wysera helps: Wyse drafts each follow-up from the note you captured, and you approve it before it sends.

For the rest of the calendar, including business and regional events, browse the AI events calendar.