Danny Bickson and Amir Alush on building tools for managing, analyzing, and unlocking visual data.
Subscribe: Apple • Android • Spotify • Stitcher • Google • AntennaPod • RSS.
Danny Bickson and Amir Alush are the creators of fastdup, a very impressive free tool for surfacing duplicates, anomalies, and leakage in visual data. In line with its name, it’s fast: fastdup is written in C++ and can handle millions of images easily. Using fastdup from Python is as simple as typing a single command. In this episode we discuss the state of tools for managing and working with computer vision data, plus the current and near-term roadmap of fastdup.
Highlights in the video version:
- Introduction to Danny Bickson and Amir Alush
State of computer vision tools and models
Key findings from their survey of computer vision teams
Describe fastdup and its core features
Accessing fastdup and capabilities
Next wave of capabilities
Computer vision and object detection
Vector databases, vector searches, and data centric AI
What’s next for fastdup? Who should try it and what to expect?
Lack of image standardization
Fastdup usage across several industries
Related content:
- A video version of this conversation is available on our YouTube channel.
- fastdup: Introducing a new free tool for curating image datasets at scale
- The Graph Intelligence Stack
- Mark Chen: How DALL·E works
- Hilary Mason: Narrative AI
- The Data Integration Market
- Jack Clark: The 2022 AI Index
- Tech companies are gearing up for the Metaverse