Machine vision for paint finish inspection

An automotive manufacturer automated paint finish inspection by connecting high-definition cameras over a private 5G network to edge compute. High-resolution video was analysed against an AI model to detect paint quality issues before vehicles left the production line.
Inspection efficiency
Overall paint inspection efficiency improved.
Detection accuracy
Paint errors identified more accurately.
Recall exposure
Fewer vehicles recalled for paint issues.
Industrial reuse
Platform created for future use cases.
Quality specialist examining the reflective paint finish of a completed vehicle at the final inspection stage of an automotive production line
Machine vision for paint finish inspection

A connected inspection environment designed to keep high-resolution image traffic moving and quality decisions close to the finishing line.

Every vehicle needed its paint finish checked before release, at the pace of high-volume production.
Objectives
The initiative set out to automate a critical quality-control step in vehicle manufacturing. The client wanted to improve the efficiency of paint inspection and reduce the human error associated with manual review, while maintaining the responsiveness needed to identify quality issues during production rather than after vehicles had left the factory.
Opportunity
The deployed private 5G network created a foundation for additional industrial use cases beyond paint inspection. Its value for future applications rests on the same capabilities required here: reliable coverage around production assets, protected traffic over the wireless interface, bandwidth for data-intensive workloads and local processing for time-sensitive decisions. The source does not identify which use cases would follow or whether any expansion has been approved.
Bringing consistent visual quality control into the flow of production

Paint inspection sits at a demanding point in automotive production. The solution needed to examine every vehicle before release, move high-resolution video through a metal-intensive factory environment and support decisions without interrupting the finishing line.

Key Challenges
1. Inspection at production volume:
The manufacturer produced a high volume of vehicles each day, and each vehicle required paint finish inspection before release.
Gloved quality inspector examining a small surface imperfection in dark blue metallic automotive paint under angled inspection lighting
2. Consistent defect detection:
The initiative needed an automated method capable of identifying several paint quality issues while reducing the exposure to human error in manual inspection.
Industrial camera mounted among metal structures and finishing equipment inside an automotive paint inspection environment
3. High-resolution video in an RF-hostile area:
Camera placement in the finishing line required wireless coverage in a radio-frequency-hostile location and enough bandwidth for high-resolution video.
Automotive inspection line with distributed camera stations, private cellular radio units and local edge equipment supporting continuous operation
4. Continuous, time-sensitive operation:
The inspection flow required responsive edge processing, protected video traffic and resilient coverage so that connectivity would not become a weak point in production.
Solution
Private 5G and edge AI for automated paint inspection

NTT deployed a private 5G network in the paint inspection section of the finishing line. The network connected high-definition cameras to an edge compute system, where an AI model analysed the video for paint quality issues.

check icon
Connected camera inspection

High-definition cameras captured the paint finish as vehicles moved through the inspection section, supporting automated review within the production process.

check icon
Edge-based AI analysis

The cameras sent high-resolution video to an edge compute system, where images were analysed against an AI model to detect several paint quality issues.

check icon
Private coverage for the finishing line

The private 5G deployment provided coverage for camera placement in an RF-hostile environment and the bandwidth required by the video workload.

check icon
Protected and resilient connectivity

Enterprise slicing protected video traffic over the wireless interface, while overlapping adjacent radio cells provided coverage resilience alongside the design’s hardware and software resilience.

Impact
More dependable paint quality decisions before release

According to the source, the client improved the overall efficiency of paint inspection and detected errors with greater accuracy. The catalogue also reports fewer vehicles being recalled from dealerships because of paint issues. No baseline, measured improvement, timeframe or attribution method is provided, so these outcomes should remain qualitative until the client validates the supporting evidence.

Drag