This week’s newsletter spotlights Voxel51’s latest blogs, key resources, upcoming talks, and a list of upcoming events.
It’s Time To Celebrate! 🥳
We are celebrating 4 years of Open Source FiftyOne 🎉
Thank you to everyone in the FiftyOne community for all that you do–everything from trying out the project to filing feature requests to squashing bugs, contributing code to helping fellow developers. Open source software like this doesn’t happen without the support of an awesome community behind it. 🙌💪
Here’s a snapshot of accomplishments and milestones in the open source community:
- 8,000 GitHub stars 🌟
- 100+ open source contributors 🏁
- 3,000 community Slack members 🧡
- 26,000+ members in our AI + CV Meetups 🌐
- 10000+ LinkedIn Followers 👍
📙 Good Reads
What Is Visual AI? Going Beyond Computer Vision
Visual AI goes beyond traditional computer vision by not only processing visual data but also enabling AI systems to perceive, reason, and make informed decisions based on that data. It’s revolutionizing industries from healthcare to retail by improving efficiency, safety, and decision-making. Voxel51 accelerates this transformation by providing tools to refine visual data and models, empowering AI builders to create robust Visual AI applications.
For more details, visit Voxel51’s blog by Brian Moore on What is Visual AI?
How I Built an In-Cabin Perception Dataset
Check out the blog by Robert Wright and Allen Lee where you’ll see how quick and easy it is to build a dataset, load it into FiftyOne and FiftyOne Teams, perform active learning techniques, integrate with transformers from Hugging Face, and ultimately get some pretty interesting revelations about the data. Although this blog post features an interior dataset for an in-cabin monitoring use case, the steps involved will work for any visual AI project. It’s intended for novices and advanced readers alike.
🔎 Good Finds
Security in Data-Centric AI: A Comprehensive Approach to Surveillance
Effective machine learning models crucially depend on high-quality data inputs. The security sector, grappling with vast volumes of video data, faces the critical task of building reliable models amidst unique challenges. In this session, we dive into how ML engineers are leveling up by prioritizing data-centric workflows to enhance model efficacy. Join Daniel Gural as we unveil strategies to identify errors, analyze embeddings, and assess security models with unparalleled precision using FiftyOne.
Have You Explored the Data-Centric AI Competition on Hugging Face Yet?
Check out Harpreet Sahota 🥑’s post on Why this competition Matters!
🧗 The Challenge
Your mission, should you choose to accept it, is to curate a subset of our provided dataset that achieves two seemingly contradictory goals:
- Reduce the overall size of the dataset
- Maintain or improve the performance of a YOLOv8m object detection model
🧰 Dataset and Resources
- A dataset of 65,986 images
- 43 object classes across categories like clothing, people, transportation, and more
- FiftyOne, an open-source tool for dataset curation and analysis.
- YOLOv8m model from the Ultralytics Model Zoo
🏆 Prizes
- First Place: $1,000
- Second Place: Top Tier Swag Package
- Third Place: Mid-Tier Community Swag Package
🎙️ Good Listens
🗓️. Upcoming Events
Check out these upcoming AI, machine learning and computer vision events! View the full calendar and register for an event.
