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.
Course Outline
Introduction to CV/NLP Deployment with CANN
- Overview of the AI model lifecycle from training to deployment.
- Key performance considerations for real-time CV and NLP applications.
- Introduction to CANN SDK tools and their role in model integration.
Preparing CV and NLP Models
- Exporting models from PyTorch, TensorFlow, and MindSpore.
- Managing model inputs and outputs for image and text tasks.
- Utilizing ATC to convert models to OM format.
Deploying Inference Pipelines with AscendCL
- Executing CV/NLP inference using the AscendCL API.
- Implementing preprocessing pipelines: image resizing, tokenization, and normalization.
- Handling postprocessing: bounding boxes, classification scores, and text output.
Performance Optimization Techniques
- Profiling CV and NLP models using CANN tools.
- Reducing latency through mixed-precision and batch tuning.
- Efficiently managing memory and compute resources for streaming tasks.
Computer Vision Use Cases
- Case study: Object detection for smart surveillance.
- Case study: Visual quality inspection in manufacturing.
- Developing live video analytics pipelines on Ascend 310.
NLP Use Cases
- Case study: Sentiment analysis and intent detection.
- Case study: Document classification and summarization.
- Integrating real-time NLP with REST APIs and messaging systems.
Summary and Next Steps
Requirements
- Familiarity with deep learning techniques for computer vision or NLP.
- Proficiency in Python and AI frameworks such as TensorFlow, PyTorch, or MindSpore.
- Basic understanding of model deployment or inference workflows.
Target Audience
- Practitioners in computer vision and NLP leveraging Huawei’s Ascend platform.
- Data scientists and AI engineers developing real-time perception models.
- Developers integrating CANN pipelines within manufacturing, surveillance, or media analytics sectors.
14 Hours
