Edge AI implementation presents significant technical challenges for developers working with resource-constrained embedded systems. Infineon Technologies AG has expanded its Edge AI portfolio with the DEEPCRAFT AI Suite.
The Suite is a collection of software, tools, and solutions designed to address common integration barriers when incorporating artificial intelligence into embedded products. The suite is now available and has been optimized for Infineon’s PSOC Edge microcontrollers.
The DEEPCRAFT AI Suite aims to support application designers throughout the development process, from initial model creation to deployment on low-power hardware. The suite addresses key engineering challenges including model optimization for embedded targets, power consumption constraints, and the complexity of integrating pre-trained models into commercial products.
The expansion of Infineon’s Edge AI software ecosystem provides hardware and software solutions that include the DEEPCRAFT AI Suite with its AI Hub containing pre-trained models and audio tuning tools. This approach allows developers to either build custom models or integrate existing ones, potentially reducing development cycles.
The core components and capabilities include:
DEEPCRAFT AI Hub: Serves as a central repository for Edge AI software resources. It contains over 50 resources including open-source models, Infineon software, development tools, and implementation examples from industrial, consumer, and automotive applications. This provides developers with reference designs and practical implementation guidance.
DEEPCRAFT Studio: An end-to-end platform for developing AI and machine learning models for edge deployment. The platform now supports computer vision in addition to audio, radar, and time-series data processing. It includes tools for training and deploying computer vision models for object detection using YOLO architectures. The platform is available at no cost when used with Infineon hardware.
DEEPCRAFT Model Converter: Addresses the challenge of model portability by converting both proprietary and open-source models for execution on Infineon hardware. The converter supports PyTorch, TFLite, and Keras frameworks, enabling deployment on PSOC Edge and PSOC 6 MCUs without extensive manual optimization.
PSOC Edge MCUs: These Arm Cortex-M microcontrollers provide machine learning acceleration alongside security features including Edge Protect Category 4 (EPC4) with PSA Certified L2 and L4 iSE, PCI pre-certification, and secure enclave protection. A dedicated 2.5D GPU enables graphical interfaces while maintaining lower power consumption compared to traditional 3D processors. The platform supports ModusToolbox, Zephyr, and DEEPCRAFT AI Suite.
The suite includes DEEPCRAFT Voice & Audio Solutions for developing voice-controlled products. DEEPCRAFT Audio Enhancement addresses speech intelligibility challenges by removing background noise, while DEEPCRAFT Voice Assistant enables local voice interface processing without cloud connectivity requirements.
These features have been optimized for PSOC Edge architectures, supporting dual-core Cortex-M55 with Ethos-U55 up to 400 MHz and Cortex-M33 + NNLite cores up to 200 MHz. On PSOC Edge, the Voice & Audio Solutions achieve always-on listening at sub-1mW power consumption with low latency in typical room conditions. Detection rates reach 98 percent in close-talking scenarios with reduced false alarm rates.
The platform delivers 75 percent faster audio processing while consuming approximately half the energy of alternative solutions. The combination of AI acceleration and memory architecture supports multi-megabyte models such as Voice ID that exceed typical DSP capabilities. Development time for AI-based voice and audio applications can be reduced by approximately 50 percent compared to traditional approaches.
Infineon’s PSOC Edge Microcontrollers and DEEPCRAFT AI Suite are currently available. Technical documentation for DEEPCRAFT AI Suite can be accessed at https://www.infineon.com/design-resources/embedded-software/deepcraft-edge-ai-solutions. Additional information about PSOC Edge is available at https://www.infineon.com/promo/next-generation-mcu.
Leave a Reply