TaiwanManufacturing2026 – ongoingCompal Electronics
CNN visual inspection and worker-action checks on notebook production lines
Compal said AI inspection lifted notebook defect-inspection yield from about 90% to above 99.5%
Problem
Optical inspection alone had stalled at about 90% yield, and high staff turnover made yields swing.
Approach
Compal used CNN deep learning to find visible defects. It added cameras that check whether workers use the right tools and screw order, and synchronized cameras that match barcodes to carton contents to catch wrong shipments. The models run on edge devices, using Compal's own gaming notebooks with ordinary cameras. Cameras were angled to capture hands only, after workers felt watched.
Results
Takeaway
Checking worker actions as well as products targeted the cause of the yield swings. The headline yield comes from a Compal project manager and was not checked independently.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
A Business Weekly column in an AI innovation series on Compal's notebook lines. It covers CNN defect inspection, camera checks on worker actions and wrong-carton prevention, with edge devices. It gives inspection yield above 99.5% from a Compal project manager and a first-pass yield rise to about 98-99%.
Full details are in the original sources.