The computer vision platform aims to help models see, reason and act in three dimensions by publishing research, open-source models and testbeds.
Roboflow announced on September 30 the creation of Roboflow Labs, a research structure dedicated to the capabilities of models in the physical world: perceiving, reasoning and acting in three dimensions, with a focus on edge vision, robotics and code.
Three areas of work
The laboratory will publish work and train models designed to run on embedded hardware, in the tradition of RF-DETR, its real-time detector licensed under Apache 2.0 license, which also covers segmentation and exposure estimation. It will also evaluate and distribute state-of-the-art laboratory models, and build test benches such as RF100-VL or Roboflow Playground with companies, focused on detection, OCR and visual reasoning.
The San Francisco-based company, backed by GV, Craft Ventures and Y Combinator, claims millions of developers, more than half of the Fortune 100 among its customers and citations in more than 13,000 research papers.
Read the full article on the original source
Go to the source →