Vicomtech – Synthetic Data Generation Using AI for the Implementation of an Assembly Assistance System
Vicomtech
Sector: Industry
Business Case
Design and implementation of a system to assist operators during the assembly process of electronic components in assembly lines, using augmented reality.
Objectives
Develop a system capable of automatically recognising electronic components in real time and projecting, by means of augmented reality, instructions to assist in their assembly.
Use case
Integration of a flexible mechanism for synthetic and automatic image generation using AI to fill the data gap for image recognition and verification processes, together with computer vision systems for object tracking.
Infrastructure
Cloud
Technology
Machine learning and deep learning Computer vision
Data
Images of real components, as well as synthetic images generated by artificial intelligence.
Resources
1 project manager, 1 HW technician (cameras and computing) and 2 SW technicians (computer vision, artificial intelligence and augmented reality). Technical development by Vicomtech, and technical validation and requirements by IKOR.
Difficulties and learning
The great difficulty is determined by the large number of electronic components available in the Ikor catalogue, as well as by the situations (boards) in which they can be deployed. The implementation of a component identification system using a combination of classifiers proved to be ideal to overcome the difficulty of the large catalogue.
KPIs (business impact and metrics of the model)
Precision/recall (detection) Frame rate Tracking quality Image generation times
Funding
The project was funded by the Basque Industry programme.
Collaborators, Partners
The project was developed for Ikor.
