It was a great time to reconnect with the research community in person, as well as our labeling partners and customers.
We were able to secure the conference’s entrance booth, meaning all of the 5,400 attendees of the conference would eventually walk by our booth. Many stopped by to talk about training data challenges, grab a cup of coffee, or pick up some branded swag.
This year’s hottest research areas had shifted slightly to the first CVPR we attended in 2019. Here are the highlights:
NERFS are going to change computer graphics
Diffusion models > GANs
New CNN architectures not making the difference, large datasets are.
Few-shot models are becoming more robust, so better "Auto Annotate" experiences are around the corner.
Transformers are making their way into industry cases.
Annotation types are becoming more uniform as research capitalizes on open-source datasets.
All in all our experience of this year’s CVPR was great—
Thanks to our booth setup and design we really stood out and attracted the attention of the attendees.
Our swag’s popularity exceeded our wildest expectations. You could see some researchers wearing V7-branded stuff. Our MNIST socks made their way to the official CVPR Twitter channel.
The party we hosted at our villa was a highlight for many given its non-corporate appeal.
The enthusiasm our team brought to the conference was infectious. We experienced a lot of positivity from everyone we talked to.
Other than ML practitioners, we brought members of our engineering, sales, and marketing team who learned a lot about upcoming computer vision trends and the challenges our users experience in research.
Last but not least, our participation in CVPR coincided with the launch of the long-awaited 2.0 features and our Free Education Plan.
And here are some pictures that tell the rest of the story: