EDGE-AI-STUDIO
適用於邊緣 AI 的軟體開發工具
EDGE-AI-STUDIO
概覽
Edge AI Studio 是一系列圖形和命令行工具,旨在加速 TI 處理器、微控制器和雷達感測器上的邊緣 AI 開發。不管是使用來自 TI Model Zoo 的模型,還是
運用您自己的模型來開發概念驗證,Edge AI Studio 都能提供您所需的工具。
Edge AI Studio 圖形使用者介面提供完全整合的解決方案,用於收集和註釋資料、訓練和編譯模型,以在即時開發平台上部署。從 TI Model Zoo 中的各種預先訓練的模型中選擇,並可選擇透過自訂資料重新訓練,以提升準確度和性能。Edge AI Studio 可作為基於雲端的應用程式供處理器裝置使用,也可提供雲端與桌面應用程式供微控制器使用。使用微控制器時,Code Composer Studio™ 開發環境會消耗 Edge AI Studio 的輸出。
對於進階使用者或使用 PyTorch 或 TensorFlow 等開放原始碼環境的使用者,TI 提供一組命令行工具,供在這些環境中使用 TI 裝置。Tensor 虛擬機 (TVM) 編譯器架構可用於最佳化 TI 裝置的模型,無論是否具有專用加速器。使用具有 C7x/MMA 等專用 AI 加速器的 TI 處理器 (MPU) 時,TI 深度學習資料庫 (TIDL) 內建了經過高度優化的通用層實現,以確保發揮最高性能。使用命令列工具時,
也可以使用自己的模型 (BYOM)。
特點
TI 支援多類應用中的邊緣 AI 解決方案開發,其應用範圍包括但不限於:
處理器 (MPU)
- 物體偵測
- 影像分類
- 語意分割
微控制器 (MCU)
- 電弧故障偵測
- 馬達故障偵測
- 鼓風機不平衡
- 一般時間序列
雷達感測器
- 表面分類
下載項目
EDGE-AI-STUDIO — Edge AI Studio
支援產品和硬體
產品
多媒體與工業網路 SoC
汽車駕駛輔助 SoC
硬體開發
開發板
EDGE-AI-STUDIO — Edge AI Studio
產品
多媒體與工業網路 SoC
汽車駕駛輔助 SoC
硬體開發
開發板
版本資訊
Model Composer is a fully integrated no-code solution for training and compiling models for deployment into edge AI applications. It supports Bring-Your-Own-Data (BYOD), enabling the re-training of models from the TI Model Zoo with custom data. Deploy to your local development board.
Tasks supported:
- Image classification
- Object detection
- Semantic segmentation
- AC/DC Arc fault detection
- Passive Infra Red (PIR) classification for CC27xx
- Motor bearing fault detection
- Blower Imbalance detection
- Time-series Classification task support for C2000, MSPM0, CC27xx
- Surface Classification for mmWave
- Point Cloud Classification task support for mmWave
最新功能
- Arc Fault Task supports AC arc fault use case on MSPM0
- Passive Infra Red (PIR) classification for CC27xx
- Time-series Classification task support for C2000, MSPM0, CC27xx
- Point Cloud Classification task support for mmWave
- Project create workflow improvements.
EDGE-AI-STUDIO-MCU — Edge AI Studio for Microcontrollers
支援產品和硬體
產品
Automotive body & lighting MCUs
Automotive powertrain, zonal & chassis control MCUs
General-purpose MCUs
Real-time digital power MCUs
Real-time motor control & automation MCUs
Sub-1 GHz 無線 MCU
Wi-Fi 產品
低耗電 2.4GHz 產品
硬體開發
子卡
開發套件
開發板
EDGE-AI-STUDIO-MCU — Edge AI Studio for Microcontrollers
產品
Automotive body & lighting MCUs
Automotive powertrain, zonal & chassis control MCUs
General-purpose MCUs
Real-time digital power MCUs
Real-time motor control & automation MCUs
Sub-1 GHz 無線 MCU
Wi-Fi 產品
低耗電 2.4GHz 產品
硬體開發
子卡
開發套件
開發板
版本資訊
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
最新功能
- Introduce additional Time-series AI tasks
- Regression and Forecasting
- Enable core AI support across TI's MCU device families
- Enabled for all AI accelerated devices and other non-accelerated devices
- Expand core AI support to AM13, AM26, CC13xx, CC27xx, CC35xx, C28xx, C29xx, MSPM0, MSPM33
- Add Sensor BoosterPack™ support for data collection, live preview across MCU LaunchPads™
- Accelerometer + gyroscope, PIR, microphone, ambient light, humidity and temperature
- Enable Library of example projects with predefined datasets
- Enhanced user experience (Revamped workspace, new project, training and deployment workflows)