![](https://csdnimg.cn/release/download_crawler_static/87919292/bg4.jpg)
Abstract
ModelArts is a one-stop AI platform for developers, providing massive data
pre-processing and interactive intelligent annotation, large-scale distributed training,
automated model generation, and end-side-cloud model on-demand deployment
capabilities for machine learning and deep learning, helping users to quickly create
and deploy models and manage full-cycle AI workflows.
Sense MindSpore is a full-scene deep learning framework designed to achieve
the three goals of easy development, efficient execution, and full-scene coverage,
providing tensor-microprogrammable capabilities that support heterogeneous
acceleration and multiple hardware platforms for cloud, server, edge, and end.
This experiment is based on Huawei Cloud ModelArts and Elastic Cloud Server
ECS to complete the whole process of AI development from training to deployment,
introducing how to build U-Net network models using the MindSpore framework,
train on the simulated data set of industrial quality inspection using online Rising
Surge computing power, and compile the saved models to generate offline models
adapted to Rising AI processors, and use the MindX SDK mxVision for inference to
achieve the task of image segmentation.
Keywords: neural network; U-Net network; Image segmentation;