System Outline

“AI×Inspection” is a system that detects the wormholes of the vegetable leaf by AI image analysis function and outputs the results as errors. The identification rate has been improved by annotating various wormholes and make them learned by AI.

Features of “AI×Inspection”

By letting AI learn the elements you want to recognize in advance, it is possible to identify and count the number of those elements from the image. Also, non-defective products and defective products can be identified by learning the elements using error values such as defects and tears. With the image identification function, error values due to temperature differences can also be detected using a thermo camera.

■ Automate inspections that relies on human eyes such as appearance inspection.
■ Detect specific objects from the image.
■ Identify types, positions and sizes.
■ By applying transfer learning, construct models with less learning data.
■ Adapt for appearance inspection and quantity counting of products.