AI Powered Wood Quality Detection
Are you sourcing real, quality timber? Due to much much lower costs, most of the furniture you buy from IKEA are NOT made out of real wood. If you value quality and long lasting wood products, how do you know if you are not being scammed by interior designers, retail shops or wholesalers?
We’re going to deep dive into a case study that combines traditional industry with high technology.
We’ve collected over 40 years of data in plywood defects from Royce Enterprise. Royce Enterprise is a 50 years furniture manufacturing business (clients include Amazon, Walmart, etc) , making us the best AI model for plywood defect detection in the world. Combined with the recent advancements in generative AI, we are able to combat timber authenticity issues.
First of all, we collect images on different types of timber logs and wood: from oak to walnut to hinkoi.
Secondly, we collect data on features that would make wood real. Fortunately, wood can be analyzed using the human naked eye, so natural features such as rims, color, grain arrangements, and swelling can be fed in as machine learning data. While artificial defects such as cross-grains, knots, and shakes are also fed into the AI model.
YOLO Model for Live Detection
For vendors who require instant defect detection for mass automated defect detection, we can employ the decade old YOLO technology.
Below is a minature convery belt that auto identifies defect wood barks using an IP camera, LED lighting and a YOLO model.
For the POC model, visit our github
