machine learning

Posted on by Jean-Luc Aufranc (CNXSoft) - 3 Comments on Pico-ITX SBC features NXP i.MX 93 LGA system-on-module

Pico-ITX SBC features NXP i.MX 93 LGA system-on-module

NXP i.MX 93 SBC

iWave Sytems iW-RainboW-G50M is an NXP i.MX 93 OSM-L compliant LGA module with up to 2GB RAM, WiFi 5 and Bluetooth 5.2 module that is found in the company’s iW-RainboW-G50S Pico-ITX SBC designed for industrial applications. The NXP i.MX 93 single and dual-core Cortex-A55 processor with an Ethos U65 microNPU was announced in November 2021, but we had yet to see any hardware based on the new NXP i.MX 9 processor family. The iW-RainboW-G50M and iW-RainboW-G50S change that with a system-on-module and single board computer. iW-RainboW-G50M NXP i.MX 93 system-on-module Specifications: SoC (one or the other) NXP i.MX 9352 dual-core Cortex-A55 processor @ up to 1.7 GHz with Arm Cortex-M33 @ 250 MHz,  0.5 TOPS NPU NXP i.MX 9351 single-core Cortex-A55 processor @ up to 1.7 GHz with Arm Cortex-M33 @ 250 MHz, 0.5 TOPS NPU NXP i.MX 9332 dual-core Cortex-A55 processor @ up to 1.7 GHz Arm Cortex-M33 @ […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 10 Comments on Achronix Speedster7t AC7t1500 FPGA is now available for high-bandwidth applications

Achronix Speedster7t AC7t1500 FPGA is now available for high-bandwidth applications

Speedster7t 7t1500 VectorPath Accelerator Card

Achronix Semiconductor has recently announced the general availability of the Speedster7t AC7t1500 FPGA designed for networking, storage, and compute (AI/ML) acceleration applications. The 7nm Speedster7t FPGA family offers PCIe Gen5 ports and GDRR6 and DDR5/DDR4 memory interfaces, delivers up to 400 Gbps on the Ethernet ports, and includes a 2D network on chip (2D NoC) that can handle 20 Tbps of total bandwidth. Achronix Speedster7t highlights: Two-dimensional network on chip (2D NoC) enabling high bandwidth data flow throughout and between the FPGA fabric and hard I/O and memory controllers and interfaces MLP (Machine Learning Processors) blocks with arrays of multipliers, adder trees, accumulators, and support for both fixed and floating-point operations, including direct support for Tensorflow’s bfloat16 format and block floating-point (BFP) format. Multiple PCIe Gen5 ports High-speed SerDes transceivers, supporting 112 Gbps PAM4 and 56 Gbps PAM4/NRZ modulation, as well as lower data rates Hard Ethernet MACs that support […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 6 Comments on ArduCam Mega – A 3MP or 5MP SPI camera for microcontrollers (Crowdfunding)

ArduCam Mega – A 3MP or 5MP SPI camera for microcontrollers (Crowdfunding)

Arducam Mega SPI camera

ArduCam Mega is a 3MP or 5MP camera specifically designed for microcontrollers with an SPI interface, and the SDK currently supports Arduino UNO and Mega2560 boards, ESP32/ESP8266 boards, Raspberry Pi Pico and other boards based on RP2040 MCU, BBC Micro:bit V2, as well as STM32 and MSP430 platform. Both cameras share many of the same specifications including their size, but the 3MP model is a fixed-focus camera, while the 5MP variant supports autofocus. Potential applications include assets monitoring, wildfire monitoring, remote meter reading, TinyML applications, and so on. ArduCam Mega specifications: Camera Type 3MP with fixed focus 5MP with auto-focus from 8cm to infinity Optical size – 1/4-inch Shutter type – Rolling Focal ratio 3MP – F2.8 5MP – F2.0 Still Resolutions 320×240, 640×480, 1280×720 x 1600 x1200x 1920 x 1080 3MP – 2048 x 1536 5MP – 2592×1944 Output formats – RGB, YUV, or JPEG Wake-up time 3MP – […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 7 Comments on T-Camera S3 – An ESP32-S3 board with camera, display, PIR motion sensor, and microphone

T-Camera S3 – An ESP32-S3 board with camera, display, PIR motion sensor, and microphone

T Camera S3

LilyGO has launched a new ESP32-S3 WiFi & BLE camera board with the T-Camera S3 also featuring a small display, a PIR motion sensor, and a microphone, as well as an optional plastic shell. The T-Camera S3 is an evolution of the TTGO T-Camera ESP32 board introduced in 2019 with many of the same features, but the ESP32 microcontroller has been replaced with an ESP32-S3 microcontroller with vector extensions that makes it suitable for machine learning and computer vision applications. The new board also comes with a larger 16 MB SPI flash, more I/Os, and a few other small changes. T-Camera S3 specifications: ESP32-S3-WROOM-1 wireless module SoC –  ESP32-S3FN16R8 dual-core Tensilica LX7 microcontroller @ 240 MHz (Note: this SKU is not listed in the official ESP32-S3 datasheet) with 2.4 GHz 802.11n WiFI 4 and Bluetooth 5.0 LE connectivity Memory – 8MB PSRAM Storage – 16MB SPI flash Camera – 2MP […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 10 Comments on Alif Ensemble Cortex-A32 & Cortex-M55 chips feature Ethos-U55 AI accelerator

Alif Ensemble Cortex-A32 & Cortex-M55 chips feature Ethos-U55 AI accelerator

Alif E7 processor block diagram

Alif Semiconductor’s Ensemble is a family of processors and microcontrollers based on Arm Cortex-A32 and/or Cortex-M55 cores, one or two Ethos-U55 AI accelerators, and plenty of I/Os and peripherals. Four versions are available as follows: Alif E1 single-core MCU with one Cortex-M55 core @ 160 MHz, one Ethos U55 microNPU with 128 MAC/c Alif E3 dual-core MCU with one Cortex-M55 core @ 400 MHz, one Cortex-M55 core @ 160 MHz, one Ethos U55 with 256 MAC/c, one Ethos U55 with 128MAC/c Alif E5 triple-core fusion processor with one Cortex-A32 cores @ 800 MHz, one Cortex-M55 core @ 400 MHz, one Cortex-M55 core @ 160 MHz, one Ethos U55 with 256 MAC/c, one Ethos U55 with 128MAC/c Alif E7 quad-core fusion processor with two Cortex-A32 cores @ 800 MHz, one Cortex-M55 core @ 400 MHz, one Cortex-M55 core @ 160 MHz, one Ethos U55 with 256 MAC/c, one Ethos U55 with […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 6 Comments on Bee Motion S3 – An ESP32-S3 board with a PIR motion sensor (Crowdfunding)

Bee Motion S3 – An ESP32-S3 board with a PIR motion sensor (Crowdfunding)

ESP32 S3 board PIR Motion Sensor

The Bee Motion S3 is an ESP32-S3 WiFi and Bluetooth IoT board with a PIR motion sensor beside the more usual I/Os, Qwiic connector, USB-C port, and LiPo battery support. It is at least the third PIR motion wireless board from Smart Bee Designs, as the company previously introduced the ESP32-S2 powered Bee Motion board and the ultra-small Bee Motion Mini with an ESP32-C3 SoC. The new Bee Motion S3 adds a few more I/Os, a light sensor, and the ESP32-S3’s AI vector extensions could potentially be used for faster and/or lower-power TinyML processing. Bee Motion S3 specifications: Wireless module – Espressif Systems ESP32-S3-MINI-1 module (PDF datasheet) based on ESP32-S3 dual-core Xtensa LX7 microcontroller with 512KB SRAM, 384KB ROM, WiFi 4 and Bluetooth 5.0 connectivity, and equipped with 8MB of QSPI flash and a PCB antenna USB – 1x USB Type-C port for power and programming Sensors PIR sensor S16-L221D […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 2 Comments on Hand Gesture Recognition on ESP32-S3 with the ESP-DL library

Hand Gesture Recognition on ESP32-S3 with the ESP-DL library

ESP32-S3 AI Gesture Recognition

Ali Hassan Shah has deployed a deep learning model for hand gesture recognition on the ESP32-S3-EYE  board using the ESP-DL library and achieved AI-powered hand recognition with a 0.7-second latency on the ESP32-S3 camera board. Last year, Espressif released the ESP-DL library for the ESP32-S3 microcontroller with AI vector extensions, as well as ESP32 and ESP32-S2, along with a face detection demo that ran much faster on the ESP32-S3. Ali rolled out his own solution for AI gesture recognition and provided a step-by-step tutorial along the way. The main steps to deploying a custom model with the ESP-DL library include: Model Development that involves Getting or creating datasets. In this case, downloaded from Kaggle with 6 gestures namely Palm, I, Thumb, Index, Ok, and C. Testing, training, and calibrating the datasets Building a (CNN) Model Training a Model Saving a Model to the Hierarchical Data format (.h5) Converting the H.5 […]

Posted on by Jean-Luc Aufranc (CNXSoft) - 20 Comments on Sipeed M1s & M0sense – Low-cost BL808 & BL702 based AI modules (Crowdfunding)

Sipeed M1s & M0sense – Low-cost BL808 & BL702 based AI modules (Crowdfunding)

Sipeed M1s & M0Sense

Sipeed has launched the M1s and M0Sense AI modules. Designed for AIoT application, the Sipeed M1s is based on the Bouffalo Lab BL808 32-bit/64-bit RISC-V wireless SoC with WiFi, Bluetooth, and an 802.15.4 radio for Zigbee support, as well as the BLAI-100 (Bouffalo Lab AI engine) NPU for video/audio detection and/or recognition. The Sipeed M0Sense targets TinyML applications with the Bouffa Lab BL702 32-bit microcontroller also offering WiFi, BLE, and Zigbee connectivity. Sipeed M1s AIoT module The Sipeed M1S is an update to the Kendryte K210-powered Sipeed M1 introduced several years ago. Sipeed M1s module specifications: SoC – Bouffalo Lab BL808 with CPU Alibaba T-head C906 64-bit RISC-V (RV64GCV+) core @ 480MHz Alibaba T-head E907 32-bit RISC-V (RV32GCP+) core @ 320MHz 32-bit RISC-V (RV32EMC) core @ 160 MHz Memory – 768KB SRAM and 64MB embedded PSRAM AI accelerator – NPU BLAI-100 (Bouffalo Lab AI engine) for video/audio detection/recognition delivering up […]