Ton bag bag opening recognition method based on ai vision

Abstract

This invention belongs to the field of artificial intelligence visual recognition and flexible packaging processing technology, specifically involving an AI vision-based method for identifying the opening of ton bags. It aims to solve the technical challenges of existing conventional vision systems, such as their inability to adapt to the flexible characteristics of ton bags, inconsistent bag opening shapes, interference from reflective materials, and changes in ambient light, resulting in difficulties in identification, low accuracy, low efficiency, and reliance on manual bagging. This method utilizes the difference between the reflective material at the ton bag opening and the bag's body material, combined with AI vision's self-learning capabilities, to achieve accurate identification of the ton bag opening, orientation determination, position positioning, and placement angle recognition. Through a reflective feature enhancement algorithm, the method highlights the reflective characteristics of the bag opening and suppresses interference from changes in ambient light and flexible deformation. A self-learning recognition model based on an improved CNN is constructed. Through repeated learning from multiple scenarios and bag opening samples with various deformations, stable identification of bag openings with different flexible deformations can be achieved without high-precision hardware support.

Images