The Age of AI Has Arrived: Insights, Trends, and Highlights from CVPR 2023

V7 leaders descended to CVPR 2023, an international conference well-known for fielding some of the most exciting reveals in AI. We address key event findings, trends, and predictions set to impact the commercial application of AI.
Read time
June 30, 2023

Whilst it's important for executives to understand future developments, even the most forward-thinking leaders could not have predicted the rate of change in AI in 2023. However, each year, the best researchers, leaders, and developers descend on CVPR, to glean insights from an international conference that has acted as the birthplace of countless AI innovations.

CVPR 2023 welcomed scores of attendees and sifted through over 9,000 paper submissions, each outlining their vision for the future of AI. Similarly, vanguards of AI were in attendance this year, with appearances from Tesla, Google, AWS, and more. Household names and AI greats revealed impressive technological advancements, increasingly closing the gap between academic endeavor and commercial reality. 

AI enters the commercial mainstream

This commercial leaning reflects much of the current climate - or rather - appetite - for AI innovation. According to a study from Accenture, 73% of companies in 2023 are prioritizing AI over all other digital investments, with 90% of business leaders applying AI to tackle aspects of operational resilience. Further, enterprise businesses are the most likely to embrace AI, with 68% of large businesses adopting at least one AI technology in 2022. 

The sprawling impacts of this shift are hard to convey, with the use of AI set to raise global GDP by 7%, according to Goldman Sachs. Perhaps most interestingly, 40% of these economic gains are set to come from AI-powered enhancements, according to research from PWC

So what are some of the commercial leaps showcased at this year’s CVPR? Discussions ranged from evolutions in generative AI, and challenges facing foundation models, to the increasing importance of data-centric AI. In terms of real-world applications, reveals included augmented reality innovations from MetaAI, virtual shoe-fittings courtesy of Snap Inc, and award-winning service robotics from DeepBlue AI.

Building on the breakthroughs of previous years, the mood of CVPR 2023 was one of excitement, with a stream of enterprises recognizing the writing on the wall: The time to strike with AI is now. 

V7 fuels AI-powered product development

Attendees echoed this fervency for AI innovation, with many seeking infrastructure that would evolve burgeoning ideas into fully fleshed products. From end-to-end model development to MLOps advancements, the event showcased the direction the AI market is heading towards: commercial application of AI, computer vision, and machine learning. 

Despite this, it’s worth noting that many of the most exciting reveals continue to come from research and innovation projects, heralding advancements in 3D labelling, diffusion models, and multimodal AI. 

Driving practical application, was V7, with a cohort of product experts in attendance. Led by the V7 founding team, we revealed a number of exciting new product releases, including SAM-powered Auto-Label features, new and improved workflows, and a new properties system for foundation model labelling.

Face-to-face with the builders of the next era of AI, use cases spanned from microfluidics, to medical imaging, from autonomous vehicles to maritime robotics, and from dental processes to agritech solutions. In contrast to the exciting fast-paced evolutions within research and innovation, these real-world applications shy away from reinventing the wheel, and rather, see AI as an undeniable cog in their mission to impactfully advance their industry today. 

Trends from CVPR attendees and predictions from V7’s leaders

So, what will the remainder of 2023 bring? What is the focus of this year’s CVPR attendees? And what are our predictions for AI-powered products?

Touching on some of the biggest talking points of 2023 - we gauged the priorities of this year’s CVPR attendees, tackling SAM, de-risking AI, and key challenges facing researchers in the coming years. 

Starting with SAM, just under half of V7 booth attendees state they have embraced the Segment Anything Model, with just under a third indicating they intend to use the model in the future. 

Pro Tip: Did you know? With the Auto-Annotate tool powered by SAM, you can pre-process whole images/videos and create semantic masks based on Meta's Segment Anything Model to enhance your ML training data tasks.

When it comes to developing safer AI solutions, respondents prioritized enhancing data quality and leveraging human-in-the-loop quality control, with the remainder highlighting a need for comprehensive governance controls and strengthening AI systems through RLHF. 

Finally, this year’s attendees tackled the key challenges AI researchers and practitioners will need to address in the coming years. 

In terms of popular research, there were a number of stand-out themes:

  • Generative AI and LLMs overall continue to dominate, with researchers clambering to address issue points like hallucinations and amplifications of bias. 
  • High-quality datasets remain the bane and need of researchers and practitioners alike, with a particular need in 3D vision and robotics. In response, simulations are on the rise, as a cheaper and more scalable alternative to real-world training. To maintain quality control, RLHF will become increasingly important, while fine-tuning could happen in real-world environments. 

Finally, what are the predictions of V7 experts, having had the opportunity to dive deep into AI advancements courtesy of CVPR? 

"Demonstrating V7 to the computer vision community at CVPR 2023 was a great experience. We were very happy to demo all of our latest and greatest features, such as our SAM integration which performs instance segmentation from a single click, AutoLabel which magically finds similar instances of a particular class, and semantic segmentation using pixel-perfect masks.

However, the highlight was the many great discussions with exciting companies, diving head-first into AI-powered product development in 2023. The appetite for AI-powered product development was immediately clear, with countless companies realizing the time to strike with AI is now. The hard work of the academic community is now paving the way for commercial application - and it’s phenomenally exciting to sit at its center."

James Petts, Medical Engineering Manager, V7

Three Big Takeaways

While CVPR 2023 was packed with a host of innovations, reveals, and trends, there were three key takeaways from the event:

  • First up, explainable AI. As a hot topic, particularly in the commercial application of AI, explainability was a key sticking point for this year’s speakers and exhibitors. Fortunately, solutions are quickly coming to the fore, not least V7’s Darwin, which has training data explainability built-in via annotation authorization, tracking, and easily governable AI infrastructure.
  • Next up, RLHF. While media reports threaten an entirely AI future, a more nuanced AI-human relationship is becoming clear. Human in-the-loop components remain a key part of the AI development process, with platforms like V7 ensuring human actions are more meaningful, more optimized, and more efficient.
  • Finally, safe AI development was a key point for CVPR 2023. This echoes the findings from IBM’s AI in the Enterprise report, which highlighted that 49% of leaders were considering delaying AI development until ethical and risk factors were legislated for. Having witnessed the monumental efforts to de-risk AI at this year’s event, we expect that statistic to dramatically shrink in 2024. 

Are you preparing to dive head-first into AI-powered product development? V7 can fuel your project, with infrastructure that: 

  • Minimizes costs throughout the AI development lifecycle.
  • Optimizes the labelling process with annotation features that prioritize speed and accuracy. 
  • Prioritizes data explainability and good governance structures.
  • Tracks and visualizes your organization’s performance, with UI that makes it simple to tweak, improve, and streamline your development process.
  • Integrates with your project within a wider AI ecosystem, with safety in mind.
  • And much, much more… 

Get in touch with the V7 team. 

Automate repetitive tasks with V7's new Gen AI tool
Explore V7 Go

Other news

Ready to get started?
Try our trial or talk to one of our experts.