For the use of artificial intelligence (AI) in production, the collection of the training data that is often required in large quantities represents a challenge. This also applies in particular to so-called computer vision solutions, with the help of which images can be processed automatically and, for example, quality inspection in production can be supported on the basis of intelligent object recognition. For the training data required for this, the states to be recognized must usually first be documented and labeled photographically in an elaborate manner.
In order to support other companies in this challenge, BMW, Microsoft, NVIDIA and idealworks have published what they say is the largest reference dataset for artificial intelligence in the field of manufacturing. The Synthetic Object Recognition Dataset for Industries (SORDI) includes a total of more than 800,000 photorealistic images of manufacturing resources, which are divided into 80 classes (e.g., forklifts, pallet cages, pallets, etc.). In addition to the images, the partners also provide associated, integrated labels as well as a no-code AI.
Accessibility
Language
