Control each metal box by artificial vision and automatically remove defects

What was done

According to Agro Media (15 June 2026), the inspection was previously based on visual controls, which were more difficult to maintain when the paces increased. The system, developed with the French company Psycle Research, analyses each box on the line and has been trained to recognize impacts, deformations and ovalisations, which can compromise crimping and sealing. A non-compliant box is ejected without stopping production. Without IA's prior experience of vision, the company first defined objectives, expected performance and detection criteria, in close connection between production teams and provider. Early removal of defective boxes has reduced fills and line stops, and an interface provides access to control data. The provider announces a performance rate of 99.97%; Babynov plans to automate other checkpoints.

Before you start

Tool cited by the source: Vision industrielle IA. Gather the information the prompt asks for. Strip confidential data that is not needed and check your tool’s access terms. Features and plans can change.

The method, in three steps.

1. 01

   Gather the context

   Gather the information described in the prompt.

2. 02

   Adapt the prompt

   Copy the prompt below, replace the bracketed items, then paste it into the AI assistant you use. The cited tool (Vision industrielle IA) is not necessarily available to the public..

3. 03

   Reread before using

   Reread the result, correct it if needed and keep the final decision. Caveats are detailed below.

Your starting point.

Replace the bracketed items, then adapt the proposal to your situation. This prompt is written by the iadujour desk It is not a quotation from the source.

I want to automate visual control on my production line.

Product and packaging:
Deficiencies to be detected:
Cadence:

1. Writes a specification for an industrial vision provider (detection criteria, acceptable rates of false releases and missed defects, integration, ejection, data).
2. Proposed to pre-production test protocol.
3. List the questions to ask the claimant.

What deserves your attention