Ascend AI-Based Moderate Rice Processing Solution
My team and I proposed an Ascend AI-based solution for intelligent rice processing. It combines Ascend computing, MindSpore, and edge-cloud heterogeneous collaboration to inspect rice quality during production and support more stable, moderate processing.
The project targets small, adhesive grain features such as rice embryos and achieves 98.5% segmentation accuracy, while detecting quality indicators including whiteness, embryo retention, chalkiness, and embryo integrity.