EDGE-AI-STUDIO
에지 AI용 소프트웨어 개발 툴
EDGE-AI-STUDIO
개요
Edge AI Studio is part of the CCStudio™ development tool ecosystem. Edge AI Studio is a collection of graphical and command line tools designed to accelerate edge AI development on TI processors, microcontrollers, connectivity devices and radar sensors. It supports both AI-accelerated devices such as processors featuring a C7™ NPU or microcontrollers featuring a TinyEngine™ as well as AI-supported devices without a dedicated accelerator.
Whether developing a proof of concept using a model from the TI Model Zoo or leveraging your own model, Edge AI Studio provides the tooling you need.
The Edge AI Studio graphical user interface provides a fully integrated solution for data management, model training and deployment on a live development platform. Select from a variety of optimized pre-trained models in the TI Model Zoo and optionally re-train them with custom data (BYOD) to improve accuracy and performance.
Edge AI Studio is available as a cloud application for processors devices and as both a cloud and desktop application for microcontrollers, connectivity devices and radar sensors.
For advanced users or those using open-source environments, such as PyTorch or TensorFlow, TI provides a set of command-line tools enabling the use of TI devices in these environments. The Tensor Virtual Machine (TVM) compiler framework is used to optimize models for TI devices, both with and without dedicated accelerators. When working with TI processors that feature a C7 NPU, the TI Deep Learning library (TIDL) contains highly optimized implementations of common layers to ensure maximum performance. With the command-line tools, it is also possible to Bring Your Own Model (BYOM).
For users of CCStudio IDE and CCStudio for VS Code the Edge AI Toolkit enables agentic integration. Use an AI agent to select, train and deploy a model to a TI microcontroller, all fully integrated into your existing IDE workflow.
특징
TI는 다음을 포함하되 이에 국한되지 않는 다양한 애플리케이션을 위한 엣지 AI 솔루션 개발을 지원합니다.
프로세서(MPU)
- 객체 탐지
- 이미지 분류
- 의미론적 세분화
마이크로컨트롤러(MCU)
- 아크 오류 감지
- 모터 오류 감지
- 블로어 불균형
- 일반 시계열
레이더 센서
- 표면 분류
다운로드
EDGE-AI-EXTENSION — Edge AI Toolkit Extension for CCStudio™ IDE and VS Code®
지원되는 제품 및 하드웨어
제품
Sub-1GHz 무선 MCU
Wi-Fi 제품
범용 MCU
산업용 mmWave 레이더 센서
실시간 디지털 전원 MCU
실시간 모터 컨트롤 및 자동화 MCU
저전력 2.4GHz 제품
차량용 무선 커넥티비티 제품
차량용 파워트레인, 구역 및 섀시 제어 MCU
차체 및 조명 MCU
하드웨어 개발
개발 키트
도터 카드
평가 보드
EDGE-AI-EXTENSION — Edge AI Toolkit Extension for CCStudio™ IDE and VS Code®
제품
Sub-1GHz 무선 MCU
Wi-Fi 제품
범용 MCU
산업용 mmWave 레이더 센서
실시간 디지털 전원 MCU
실시간 모터 컨트롤 및 자동화 MCU
저전력 2.4GHz 제품
차량용 무선 커넥티비티 제품
차량용 파워트레인, 구역 및 섀시 제어 MCU
차체 및 조명 MCU
하드웨어 개발
개발 키트
도터 카드
평가 보드
문서
출시 정보
CCStudio™ Edge AI Toolkit Extension is a VS Code extension for CCStudio™ IDE and VS Code®. It detects, installs, and updates the environment needed for TI Edge AI (TinyML) workflows: a local Python interpreter, the TinyML ModelZoo training and compilation toolchain, and the TinyML AI Skills that teach your AI coding agent the workflow.
Once installed and configured, it will provide Edge AI Agentic workflow capabilities inside CCStudio IDE or CCStudio for VS Code, including dataset management, model training and optimization, and importing trained models into your CCStudio projects, for compiling, loading, running live inference, and debugging on device. The included agent skills system provides your AI agent with knowledge of all the features and settings needed for Edge AI development on TI devices, so it can answer specific questions, offer model or device recommendations, or guide you through an entire development workflow. Try it out, and start prompting it with the Edge AI application you want to build!
새 소식
- Initial Release
- Guided setup of the Edge AI environment from a single Setup panel
- Local Python 3.10 interpreter, or your own 3.10.x interpreter / virtual environment
- TinyML ModelZoo v1.5 for dataset management, model training and optimization, and compilation
- TinyML AI Skills v1.5 that teach your AI agent the end-to-end Edge AI workflow
- Update notifications when newer ModelZoo, Skills, or Python packages are available
- Automatic AI agent configuration, so your agent can find the installed Python and ModelZoo with no manual setup
- Starter prompts to help you begin working with your AI agent
EDGE-AI-STUDIO-MCU — Edge AI Studio for Microcontrollers
지원되는 제품 및 하드웨어
제품
Sub-1GHz 무선 MCU
Wi-Fi 제품
범용 MCU
실시간 디지털 전원 MCU
실시간 모터 컨트롤 및 자동화 MCU
저전력 2.4GHz 제품
차량용 파워트레인, 구역 및 섀시 제어 MCU
차체 및 조명 MCU
하드웨어 개발
개발 키트
도터 카드
평가 보드
EDGE-AI-STUDIO-MCU — Edge AI Studio for Microcontrollers
제품
Sub-1GHz 무선 MCU
Wi-Fi 제품
범용 MCU
실시간 디지털 전원 MCU
실시간 모터 컨트롤 및 자동화 MCU
저전력 2.4GHz 제품
차량용 파워트레인, 구역 및 섀시 제어 MCU
차체 및 조명 MCU
하드웨어 개발
개발 키트
도터 카드
평가 보드
출시 정보
Edge AI Studio is part of the CCStudio™ development tool ecosystem. Edge AI Studio is a collection of graphical and command line tools designed to accelerate edge AI development on TI processors, microcontrollers, connectivity devices and radar sensors. It supports both AI-accelerated devices such as processors featuring a C7™ NPU or microcontrollers featuring a TinyEngine™ as well as AI-supported devices without a dedicated accelerator.
The Edge AI Studio graphical user interface provides a fully integrated solution for data management, model training and deployment on a live development platform. Select from a variety of optimized pre-trained models in the TI Model Zoo and optionally re-train them with custom data (BYOD) to improve accuracy and performance.
Features
- TI Model Zoo containing a wide variety of optimized models
- Supports tasks and models across Time Series, Vision, and Radar
- Supports devices across processors, microcontrollers, connectivity devices and radar sensors
- Data collection
- Data annotation
- Model training with custom data (BYOD)
- Deploy to a live device
- Integration with CCStudio™ IDE
새 소식
- New Applications (Wi-Fi CSI Presence Detection, Cough Detection, Google Speech Commands, Wake Word Detection, Glass Break Detection)
- Visualization for Wi-Fi Channel State Information (CSI) data format
- Built-in playback for WAV file format
- Dataset Format Checker: reports errors and warnings for datafiles in an incorrect format
- New Feature Extraction Pipeline GUI for new Training Runs
- Model Quantization: 2, 4, or 8 bits, or Mixed Auto Precision
- Training Comparison: select two existing training runs to compare
- Parallel Training Runs: run up to 4 training runs in parallel
- Various enhancements and bug fixes