Gestamp – Artificial Intelligence for Image-Based Weld Defect Detection
Gestamp
Sector: Industry
Business Case
Project defined to develop an algorithm that, based on weld bead images available from existing weld bead scanning systems (e.g., Mapvision, SmartRay), will be trained to classify defect type, and those defects will be correlated with process data.
Objectives
Automatic signal validation rules. Automatic image-based classification of weld defects.5. Identificar ciclos de soldadura anómalos
Use case
Use case in a Gestamp plant to be defined, but comprising at least one assembly line with an image acquisition system and at least two of the following technologies: PHD, cold stamping and ecoat, based in Spain. What is developed in this plant can be scaled up to other assembly lines and plants.
Infrastructure
TBD
Technology
TBD
Data
TBD
Resources
AI team, Gestamp adoption area and welding experts.
Difficulties and learning
TBD
KPIs (business impact and metrics of the model)
TBD
Funding
Collaborators, Partners
Tecnalia
