Industrial
Metaverse
The Convergence of Physical and Virtual Space
The gap in the state of the art
Scientific and technical progress requires comparability, yet many industrial datasets today are either locked behind proprietary barriers or purely academic in nature. To bridge this gap, the ARENA2036 real-world laboratory, in close collaboration, has developed the IRIS benchmark in close collaboration.
A benchmark for industrial perception
Our approach is deliberately different from conventional datasets:
- Increasing the Level of Difficulty: We drastically increase the difficulty for the AI by including components in the dataset that look to one another (high intraclass variance, e.g., O-rings of different sizes), which forces the use of more precise algorithms.
- Democratizing Access: We are lowering the Hbbarriers toüSMEs and researchers. Since the benchmark is based on globally collectedä, physical access to the test objects isüfor everyone and cost-effectiveütoppossible.
A Contribution to the Industrial Metaverse
IRIS is open source. We invite companies and researchers to test their algorithms against our benchmark. With SOTA (State-of-the-art) results, we set a benchmark for the quality of synthetic training data.