AI Module Technology
Empowering Edge AI with Flexibility and Efficiency
Experience seamless edge computing with diverse hardware, built-in AI models, and multi-framework compatibility. Optimized for rapid deployment and peak performance with low power consumption.
Diverse Hardware Options
- Flexible Configurations: Offers embedded modules, USB modules, and standalone solutions, adapting seamlessly to various device architectures and application scenarios.
- Plug-and-Play Support: Simplifies device integration, shortens hardware upgrade cycles, and lowers the barriers to achieving intelligent systems.
Built-in AI Model Support
- Preloaded Application Models: Ready-to-use models eliminate the need for additional development, enabling quick deployment.
- Supports a wide range of preloaded AI models: including classification models such as MobileNet, ResNet, and InceptionNet; detection models like YoloV5, SSD, and MTCNN; pose analysis models including OpenPose and MultiPose; image processing models such as UNet, EDSR, and SID; and speech analysis models like DeepSpeech and RNNoise.
Compatibility with Frameworks and Models
- Support for Popular Frameworks: Compatible with mainstream deep learning frameworks such as PyTorch, ONNX, TensorFlow, TensorFlow Lite, Keras, and Caffe, meeting diverse development needs.
- Flexible Model Portability: Enables seamless switching between new and existing models, helping developers quickly adapt to market changes.
Automated Toolchain and Streamlined Development
- AI Automated Toolchain: Automatically generates application models with training data input, simplifying the model training and deployment process.
- Comprehensive Development Environment: Equipped with user-friendly APIs, enabling rapid development while reducing software development time and labor costs.
Performance and Reliability of Edge Computing
- Dynamic Resource Allocation: Automatically optimizes the utilization of TPU, VDSP, and CPU based on workload, enabling multitasking and efficient execution.
- Localized Data Processing: Enhances application response speed and strengthens data privacy protection, meeting the high demands of edge applications.
Flexible Scalability and Sustainable Technical Support
- Compute Stacking Capability: Expands from lightweight devices to high-performance computing scenarios, meeting growing application demands.
- Ongoing Model and Algorithm Updates: Ensures long-term stability and competitiveness with continuous support for new model architectures and algorithm advancements.