Part Recognition with Computer Vision + Metadata AI
Partium’s image recognition technology can identify spare parts across many industries — engineering, railway, heat treatment, automotive, home improvement. We combine our image similarity models with feature-specific classifiers and AI-enabled metadata filtering to deliver hyper-precise matching of parts.
We offer a mobile SDK and white-label applications that can enable users to identify spare parts, find information on them, and replace them.
This appliance, with an onboard GPU and light box, enables on-device recognition of parts, training data collection, and testing, all in one heavy-duty kiosk. Perfect for stores, warehouses, and industrial settings.
Part recognition flow
1. Pre-Search Filtering
When necessary, filter based on criteria like category, worksite, warehouse, or car make/model.
2. Calibration and Image Processing
Enables on-device recognition of part size, and scale of the object, as well as barcode and print recognition (if the image is from a catalog).
3. Object Recognition
Best matching parts produced by our visual search AI models in the case an exact match is not produced from barcode or print.
4. AI-Enabled Disambiguation
In the form of intelligent questions presented to the user to distinguish non-visual features of parts, and narrow down the candidates to a single result.