Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 14 hours
Course Outline
Introduction to Edge AI and Ascend 310
- Edge AI overview: trends, constraints, and applications.
- Architecture of the Huawei Ascend 310 chip and its supported toolchain.
- The role of CANN within the edge AI deployment stack.
Model Preparation and Conversion
- Exporting trained models from TensorFlow, PyTorch, and MindSpore.
- Utilizing ATC to convert models to OM format for Ascend devices.
- Addressing unsupported ops and implementing lightweight conversion strategies.
Developing Inference Pipelines with AscendCL
- Executing OM models on Ascend 310 using the AscendCL API.
- Managing input/output preprocessing, memory handling, and device control.
- Deploying within embedded containers or lightweight runtime environments.
Optimization for Edge Constraints
- Reducing model size and tuning precision (FP16, INT8).
- Identifying bottlenecks using the CANN profiler.
- Optimizing memory layout and data streaming for improved performance.
Deploying with MindSpore Lite
- Leveraging the MindSpore Lite runtime for mobile and embedded targets.
- Comparing MindSpore Lite with the raw AscendCL pipeline.
- Packaging inference models for device-specific deployment.
Edge Deployment Scenarios and Case Studies
- Case study: Smart camera with object detection model on Ascend 310.
- Case study: Real-time classification in an IoT sensor hub.
- Monitoring and updating deployed models at the edge.
Summary and Next Steps
Requirements
- Prior experience with AI model development or deployment workflows.
- Foundational knowledge of embedded systems, Linux, and Python.
- Familiarity with deep learning frameworks such as TensorFlow or PyTorch.
Target Audience
- IoT solution developers.
- Embedded AI engineers.
- Edge system integrators and AI deployment specialists.
Testimonials (1)
That we can cover advance topic and work with real-life example
