How can AV teams stop drowning in petabytes of data and actually ship safer autonomy faster? Fortellix’s Rohan Bhasin and Voxel51’s Dan Gural explain how neural reconstruction, scenario-driven data curation, and NVIDIA-powered pipelines turn ordinary drive logs into high-fidelity simulations that close the last-mile gap in AV performance.
GTC is the premier global AI conference. Learn more at nvidia.com/gtc
Tin Sohn, technical lead for vision-language-action models at Porsche, and Brian Moore, CEO and co-founder of Voxel51, explore how AI, data, and simulation are shaping the future of autonomous vehicles. They share insights on the industry's transition from rule-based systems to data-driven, end-to-end approaches, the growing use of synthetic and simulated data for safety-critical testing, and how foundation models can enable cars to reason, act, and even interact like human drivers. Learn more at ai-podcast.nvidia.com.
Jason Corso speaks to Jon Krohn in this Five-Minute Friday all about Voxel51’s latest tool, Verified Auto-Labelling, and the company’s incredible success in developing popular tools for computer vision.
Additional materials: www.superdatascience.com/906
Interested in sponsoring a SuperDataScience Podcast episode? Email natalie@superdatascience.com for sponsorship information.
Today, we're joined by Jason Corso, co-founder of Voxel51 and professor at the University of Michigan, to explore automated labeling in computer vision. Jason introduces FiftyOne, an open-source platform for visualizing datasets, analyzing models, and improving data quality. We focus on Voxel51’s recent research report, “Zero-shot auto-labeling rivals human performance,” which demonstrates how zero-shot auto-labeling with foundation models can yield to significant cost and time savings compared to traditional human annotation. Jason explains how auto-labels, despite being "noisier" at lower confidence thresholds, can lead to better downstream model performance. We also cover Voxel51's "verified auto-labeling" approach, which utilizes a "stoplight" QA workflow (green, yellow, red light) to minimize human review. Finally, we discuss the challenges of handling decision boundary uncertainty and out-of-domain classes, the differences between synthetic data generation in vision and language domains, and the potential of agentic labeling.
The complete show notes for this episode can be found at https://twimlai.com/go/735.
0:46 Jason intros Dr. Brian Moore
3:39 What does Orreco do and how do they help elite athletes?
4:46 Do they sample blood?
6:33 What teams do they work with?
7:50 What is a bio-marker?
9:05 Why has inflammation monitoring become so popular?
15:01 Brian demos the Orreco app
22:56 What are the top reasons for elite athletes to use Orreco?
28:06 How did Brian get into this business?
30:34 How does blood differ in elite athletes? Is it regional?
36:25 What does Orreco have the biggest impact on?
43:40 What kinds of performance optimization should or shouldn't be legal in pro sports
51:12 How have they monitored machine learning?
56:47 Importance of glucose monitoring
59:15 Correlation in NBA between minutes per game and years played?
1:01:29 Taking a personal loan out to fund Orreco in the early days
1:15:39 Marc Andreessen asks Jason a question