Vision Data Collector¶

Are you searching for a standard and convenient way to collect vision data for your AI solution?¶

Vision Data Collector is a Siemens Industrial Edge App which captures images from Vision Connector, inference results from AI Inference Server, and stores them on a file server. You can use the captured data to (re)train and diagnose AI models.

Key features and benefits¶

• Key features

  • Collect images from Vision Connector via ZeroMQ
  • Collect images from the shopfloor according to defined rules
  • Store those images locally (file system)
  • Collect prediction results for those images and store them locally
  • Support image conversion and image resizing
  • Provide data for visualization on Edge via WebSocket API
  • Send collected data to the external environment (pull/push mechanism) according to the defined rules (FTP/STFP, Customer Endpoint)
  • Allows custom sync to Cloud (e.g. MDF, FlowCreator, Kafka)
  • Simple GUI for system setup, rules configuration and logs

• Benefits

  • Standard and convenient way to collect vision data for (re)training.
  • Close the MLOps loop for IAI Portfolio.
  • Enables additional rich functions with image metadata, such as inference result visualization/calibration, automatic image labeling, data filtering, etc..
  • Collect images and inference result in a configurable way.
  • Store data in local file system with customized storage volume.
  • Provide data flexibly (local export, enterprise-level server storage, to cloud, etc.).

Available APIs¶