Standard Model Deployment

ModelArts Standard provides capabilities for managing models and services, supporting unified management of images and models with varying frameworks and functions from multiple vendors.

AI model deployment and large-scale implementation are typically complex tasks.

For example, in smart transportation projects, after obtaining a trained model, it needs to be deployed in various scenarios such as cloud, edge, and device. Deploying on the device requires deploying it to cameras of different specifications and vendors, which is a time-consuming and challenging endeavor. ModelArts supports one-click deployment of trained models to various devices and scenarios, including devices, edges, and the cloud. It also provides a comprehensive and reliable one-stop deployment solution for individual developers, enterprises, and device manufacturers.

Figure 1 Process of deploying a model

• Real-time inference services enable high concurrency, low latency, elastic scalability, and support multi-model grayscale release and A/B testing.
• It supports deployment in various scenarios, including real-time and batch inference services on the cloud.