Soft packaging of milk, coffee, snacks and other foods is printed with production date, batch number, origin and other information related to food safety, which plays a very important role in preventing buyers from eating expired or counterfeit inferior goods. In the production process of these foods, due to machine errors or various reasons, this information may be missing, blurred, character position deviation etc.
Due to wrinkles and reflections on the surface of flexible packaging, defective products must be detected quickly and accurately on a high-speed production line. Traditional machine vision inspection sometimes misdetects and affects the inspection speed.
The AI visual inspection system developed by Intsoft Tech, combined with the OCR industrial camera, can achieve a 99.99% inspection success rate.
Intsoft Tech AI visual inspection system has many advantages:
1. Easy wrinkle surface character recognition
AI deep learning system combine with industrial ocr camera can accurately identify the production date, origin, batch number and other product information on the package, and has no requirements for the shape of the packaging. By training the AI, text on the wrinkled part of the surface can also be accurately recognized.
2. Eliminate reflection interference
Text and background can be separated during recognition. Combined with AI algorithms, it perfectly solves the problems of visual fatigue and difficulty in recognition caused by reflection on the packaging surface and various colors during manual inspection and traditional visual inspection.
3. Higher accuracy
The deep learning system only needs simple manual annotation training to automatically learn the defect type without the participation of professionals. With the increase of samples, the recognition accuracy can approach 100% infinitely.
4. Short development time
Deep learning is an upgrade of traditional algorithms, which solves the problems of traditional solutions such as difficult lighting and long algorithm development time, and greatly saves production costs.