← Cases
South KoreaManufacturing2026 – ongoing

Sunil Safe

AI vision inspection of safe painting under a smart-factory support program

Safe maker Sunil Safe added AI vision inspection of painting and aims to cut defects from 3% to 0.5%

In progressCompany's own claim

Problem

Every safe Sunil Safe makes goes through a painting step, and paint defects were running at about 3%. Defective units are scrapped, which adds disposal costs for a small manufacturer.

Approach

The company, founded in 1972, had already completed five rounds of smart-factory work, including computerized systems and a manufacturing execution system. Through the AI track of the government and Samsung Electronics co-operative smart-factory program, which covers 75% of project costs up to 300 million won, it installed AI vision cameras to judge paint defects and turned the step that wipes off oil left from earlier processing into a standard recipe.

Results

3% → 0.5%paint defect rate before the AI inspection, and the rate the company expects to reachCompany's own claim[1]
75%share of project costs covered by the program's AI track, up to 300 million wonPress reporting[1]

Takeaway

Earlier digitization projects supplied the process data and systems for AI. The company has not reported a measured result for its 0.5% defect-rate target.

Categories

Business function
Production and quality
Innovation type
Process innovation
AI technology
Computer vision
Evidence
Company's own claim

Sources

  1. 韓 중소기업들, 삼성전자 스마트공장 지원에 혁신·생산성 UP
    Metro Seoul (메트로신문)13 Oct 2026News article

Source summaries

[1]Metro Seoul (메트로신문)13 Oct 2026

Metro Seoul reports on small manufacturers at Smart Biz Expo 2026 that received smart-factory support with Samsung Electronics. It profiles Sunil Safe, founded in 1972, which uses AI vision cameras to judge paint defects and standardized its oil-wiping step; its research director says the company expects to cut the defect rate from about 3% to 0.5%. It states that firms in the program's AI track receive 75% of project costs, up to 300 million won.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026
Open case page ↗