How PellencST cut their time-to-model by 90% and gained a competitive edge in the sorting machine market
Transforming waste sorting machines with vision AI
Accelerating time to action
Defects Detection
How to centralize data from 20+ different plants, manage 100k+ monthly new data smoothly, and bring together 10 Computer Vision engineers across 5 countries.
MANUFACTURING
COMPUTER VISION ENGINEER
POC
Context
A leading steel-manufacturing company has a computer vision initiative and a great number of use cases involving defect detection on the production line. Quality assessment is often the primary issue in plants.
Their innovation team is spread across more than 5 R&D centers in Europe and the US, making collaboration and knowledge-sharing key factors for their success.
why choose us
1 single point of truth for every CV engineers.
+10 new models trained monthly decrease time spent by data scientists on low-value tasks.
1 month
From 6 to 1 month to train a test a new approach
Challenge
1
Complex end-to-end model process to be handled by a small team of data scientists.
2
Facilitate collaboration between people with different deep learning skills.
3
Ease the collaboration between people with different levels of Deep Learning skills.
4
Standardize POC approach to raise development standards.
Solution
Model Library: Leverage prebuilt Computer Vision models to train their first models and assess project feasibility.
Tracking & Versioning: Use Picsellia's tracking and versioning system to share trained models across teams.
SDK: Create development standards with Picsellia’s SDK so everyone works the same way.
Hybrid deployment: Store data and deploy models on their own infrastructure. Leverage the Picsellia GDPR infrastructure for model training.
FR
EN