Use Cases
Sep 22, 2025
Driving Edge AI Innovation: Synthetic Can Dataset for Manufacturing
Aluminum can manufacturing is operating at massive scale. Industry reports estimate global aluminum can output at about 627 billion units in 2024, which is roughly 1.72 billion cans per day. In North America, production was about 120 billion cans in 2020 with projections to reach 173 billion by 2030.
syntheticAIdata partnered with EDGE AI FOUNDATION to create a large scale synthetic dataset that accelerates computer vision for manufacturing. The dataset includes 6 million high quality images designed for training and validating edge ready models.
syntheticAIdata partnered with EDGE AI FOUNDATION to create a large scale synthetic dataset that accelerates computer vision for manufacturing. The dataset includes 6 million high quality images designed for training and validating edge ready models.
Customer
EDGE AI FOUNDATION
Industry
Manufacturing
Location
Los Altos, California, US
Platform
Industrial robots
Customer
The EDGE AI FOUNDATION is a vibrant global community dedicated to democratizing and accelerating advancements in edge AI technologies. It serves as a dynamic hub for knowledge exchange, networking, collaboration, advocacy, and education in edge AI. Through EDGE AI FOUNDATION Labs, the community curates practical models, blueprints, and datasets.
syntheticAIdata is a partner of EDGE AI FOUNDATION. The partnership was announced in May 2025 and supports joint efforts to advance edge computer vision for industry and manufacturing.
Website: www.edgeaifoundation.org
EDGE AI FOUNDATION Labs: edgeai.modelnova.ai
Challenge
Training edge models to detect and classify food and beverage cans requires large volumes of consistent, diverse imagery. Collecting real world data is costly and slow, and it is difficult to control lighting, camera angles, and backgrounds across production lines.
Cans come in many shapes, tab designs vary, and both tab and can colors differ across products and brands. Images of broken or bent tabs are especially scarce because quality control removes these defects quickly and capturing them at line speed is difficult.
chevron_left
chevron_right
Solution
syntheticAIdata developed a can dataset of 6 million synthetic images purpose built for manufacturing and industrial use. Designed for edge deployments, it delivers high quality, controllable imagery that accelerates training and evaluation of computer vision models at the edge.
The dataset covers metal and aluminum food and beverage cans rendered under varied lighting conditions and captured from two camera angles. Each scene is produced in both low and high resolution to align with edge hardware constraints and training needs. The dataset is available in EDGE AI FOUNDATION Labs so teams can prototype, benchmark, and iterate quickly.
Result
With 6 million realistic and diverse synthetic images accessible in EDGE AI FOUNDATION Labs, teams can train and validate models faster, improve accuracy on edge hardware, and cut reliance on costly real world data collection. This shortens time to deployment and supports reliable performance from prototype to production.
We’re excited to keep building with the EDGE AI FOUNDATION’s Datasets and Benchmarks Working Group, expanding coverage to new use cases and defect types while raising the bar for edge AI in manufacturing.
syntheticAIdata is leveraging generative AI to develop high quality datasets that will enable faster deployment of AI solutions in the real world. Their support of Datasets and Benchmarks Working group and their contributions will help the entire community accelerate.
Pete Bernard, CEO of EDGE AI FOUNDATION