An affordable SDI signal generator using Arduino MKR Vidor 4000

Transmission of video from one location to another in the professional video world is typically done via SDI (Serial Digital Interface). This FPGA-powered device, intended as a test tool for SDI circuits, pushes signals at 1.485 Gbps. Based on an Arduino MKR Vidor 4000, Chris Brown of the sports media technology company Tempus Ex has created “possibly the world’s cheapest SDI signal generator.”

“In the world of professional video, SDI (Serial Digital Interface) is the industry standard for getting video from point A to point B,” Brown provides context before explaining.

“It’s the professional’s equivalent of HDMI for consumer electronics. At any large-scale, televised event, you can bet that somewhere behind the scenes, there’s a rat’s nest where all of the videos come together and are distributed via SDI.”

The carrier board for Chris Brown’s Arduino MKR Vidor 4000 Carrier Is “Possibly the World’s Cheapest SDI Signal Generator.”

The cost of SDI hardware tends to be higher than one might expect for something designed for professionals, but Brown notes that while the standard can support high bitrates, SDI is actually quite straightforward. He created a low-cost SDI signal generator, which he refers to as “the ‘hello world'” of SDI hardware, to demonstrate his point.

“To keep it simple, we’ll just make the signal some flashing colors. And we’ll aim to make this a 1080p30 signal, which can be transmitted via HD-SDI,” Brown composes. “In theory, we just need to program a microcontroller that can send a 1.485 Gbps signal to a cable driver. This would work if we could get our hands on such a microcontroller, but 1.485 Gbps is quite fast. To be able to generate a signal at that speed, we’ll need an FPGA, which can be programmed using a lower-level hardware description language to perform much faster operations.”

The Arduino MKR Vidor 4000 has both a microcontroller and an Intel Cyclone 10 FPGA, which is exactly what is required for the project. Although its performance falls short of the level required for nearly 1.5 Gbps of data throughput, a Semtech GS2962 serializer completes the build; ten data lines, each delivering 148.5 Mbps, are run in parallel to achieve the required level of performance.

To achieve the desired level of performance, the Arduino MKR Vidor 4000 employs Cyclone 10 FPGA. (Source: Chris Brown)

“As far as I know,” Brown writes,

“this is the first time an Arduino has been capable of generating and emitting arbitrary SDI video signals, and as far as I know, this is also the cheapest SDI signal generator available anywhere. SDI signal generators typically cost anywhere from $350 to $2,000. This one costs ~$87 for the Arduino and ~$95 for the transmitter parts and assembly at low volumes. Although I don’t recommend using an Arduino like this in mission-critical workflows, it can be a valuable tool in the lab.”

The complete article by Brown is available on the Tempus Ex blog, while the design files and source code are available on GitHub under an open-source license that is not specified.

Translating “Dah” and “Dit” Speech Into Morse Code with an CWVox using Arduino

image: CWvox Prototype connected to the radio (Source: Kevin Loughin)

These days, it seems like every speech-detection-related project is based on some kind of advanced algorithm, such as machine learning, pattern matching, or spectrum analysis. But Kevin Loughin (AKA KB9RLW), an amateur radio operator, built CWvox, a voice-activated keyer that uses the analog-to-digital converter (ADC) on an Arduino Nano to translate spoken dits and dahs into morse code. Morse code, also known as “CW,” is a system of secret symbols (or words) based on alternating short and long tones. To represent the short burst in writing, a dot is used, while the long burst is represented by a dash.

Most people’s mental image of Morse code involves the short and long tones produced by a device called a keyer. The majority of these gadgets feature paddles that are activated with the user’s thumbs and forefingers. So, what do you do if you lose the ability to use your fingers and thus a keyer? CWvox is necessary due to this snag.

Those who work with radios will recognize these tones as “Dit” (short) and “Dah” (long) in Morse code. It turns out that there is a tiny difference in the amount of time it takes to say each of these words, and that difference is enough for effective and straightforward word detection. At least when your focus is limited to a single phrase!

CWvox Schematic (Source: kb9rlw.blogspot.com)

The heart of CWvox is an Arduino Nano. The Nano microcontroller is a basic eight-bit design without DSP capabilities. To translate verbal morse code into electronic tones, you only need that board, a few transistors, and a few passive parts. For the Nano’s positive-only ADC, the discrete parts on the input side amplify, bias, and filter the signal. A BJT buffers the signal on the output side before it enters the radio’s keying input.

The CWvox voice-operated keyer for versal morse code (Source: Kevin Loughin, Youtube)

In the above video, at 09:45, KB9RLW gives a live demonstration of CWvox in use. And it’s amazing to watch and listen to! To view a detailed explanation of the overall circuit, fast forward to 13:15 in the same video.

The complete CWvox article is available here. For the Arduino Nano code, you can also look at this GitHub repository. You should also look at the code to see how elegantly straightforward this design is!

Radxa’s Rock 5A, Rivals the Raspberry Pi 4

Radxa has unveiled the Rock 5A, a Raspberry Pi alternative that looks very similar to the Raspberry Pi 4, but with the added benefit of 8K video support and up to 16GB RAM. The new Rock 5A is not so different from the Raspberry Pi 4 Model B. It is roughly the same size as a Raspberry Pi 4 Model B, but with higher performance.

The Rock 5A is an attractive option compared to the Raspberry Pi due to its price and performance. For only $99, you can pre-order the 4GB model, which has enough RAM for most single-board computer users. Retro gamers will likely pick this model as it has enough RAM to emulate classic games, including some from the PlayStation 2 era, which is now considered retro.

Rock 5A backside

The Rockchip RK3588S system-on-chip (SoC) is the foundation of the new board, featuring four Arm Cortex-A76 cores running at 2.2-2.4GHz and four lower-power Cortex-A55 cores running at 1.8GHz. Additionally, it has an ARm Mali G610 MC4 graphics processor with OpenGL ES 3.2, OpenCL 2.2, and Vulkan 1.2 support, a neural processing unit (NPU) accelerator with a claimed to compute performance of 6 trillion operations per second (TOPS) at INT4 precision, and a selection of 4GB, 8GB, or 16GB of LPDDR4x memory modules.

The Rock 5A features two micro-HDMI video outputs. One of the outputs supports 8k60 and the other 4k60 output, as well as a four-lane MIPI Display Serial Interface (DSI) connector, all of which can be used simultaneously. It also enables hardware video decoding of H.265/H.264/AV1/AVS2 at up to 8k60, a single physical MIPI Camera Serial Interface (CSI) that can function as one four-lane or two two-lane ports, an eMMC connector, and an M.2 E-key slot for high-speed storage, a microSD slot, two USB 2.0 ports, two USB 3.0 ports (one of which can act as a USB On-The-Go (OTG) port), and a gigabit Ethernet port.

No wireless networking is available on the board however, it can be added with a USB dongle or an M.2 card. It does offer a real-time clock with optional battery backup, a user-controllable status LED, a fan connector with pulse-width modulated (PWM) speed control, and a 40-pin general-purpose input/output (GPIO) header with a range of signals including one CAN bus, up to three SPI and six I2C buses, two SPDIF and one PCM/I2S signals, up to five UARTs of which two have hardware flow control, seven PWM pins, one analog input, and two each of 5V and 3.3V power pins.

Radxa has announced that the board will be available for purchase at the end of Q1 2023, with the 4GB model priced at $99, the 8GB model at $119, and the 16GB model at $159. However, through a partnership with ALLNET China, pre-order codes can be purchased for $5, which will provide a $30 discount, reducing the prices to $74, $94, and $134 respectively. For more information, visit the Radxa forum page.

Variscite Launches New System on Module Powered by Texas Instruments’ AM62x from only $33

Variscite, a leading worldwide System on Module (SoM) designer, developer, and manufacturer, today announced the launch of the VAR-SOM-AM62, powered by the Texas Instruments AM62x. Starting from only $33, the module provides an ideal solution for embedded industrial products that require high-performance and low-power.

The VAR-SOM-AM62 runs on 1.4 GHz Quad-core Cortex™-A53 with 400MHz Cortex-M4F an additional 333 MHz PRU real-time co-processors. It offers camera interfaces MIPI-CSI2, integrated 3D GPU along with dual LVDS display. The advanced feature set complements rich connectivity with certified dual-band Wi-Fi, BT /BLE 5.2, 3x CAN bus, dual USB, and dual GbE, and supports industrial operating temperatures.

Pin compatibility and Scalability

The SoM is a member of Variscite’s Pin2Pin product family, allowing developers to use the same carrier board design for a wide range of processors. Variscite customers enjoy significant long-term advantages as well as reduced development time, costs, and risks.

“The integration of the AM625x processor into the VAR-SOM-AM62 gives developers a solid foundation that combines performance and power efficiencies,” said Ofer Austerlitz, VP Business Development and Sales of Variscite. “These qualities along with its integrated interfaces generate a very attractive cost-optimized solution for present-day embedded edge devices.”

VAR-SOM-AM62 Evaluation kits

The VAR-SOM-AM62 evaluation kits serve as a complete development platform for both evaluation and application development purposes and include the scalable Symphony carrier board and an optional LVDS display with a touch panel.

Availability and Longevity

The VAR-SOM-AM62 is included in Variscite’s long-term longevity plan and provides longevity of 15 years. Evaluation kits and stock item modules are available for production quantities orders, starting from only $33 per unit.

ESP32-P4 Offers High-Performance MCU with Numerous IO-Connectivity

Espressif Systems has unveiled its newest SoC, ESP32-P4, which is powered by a dual-core RISC-V CPU with an AI instructions extension, an advanced memory subsystem, and integrated high-speed peripherals. ESP32-P4 is designed to meet the needs of high-performance applications that require strong security, as well as the next generation of embedded applications that will require robust support for Human-Machine Interfaces, efficient edge computing, and increased IO-connectivity.

ESP32-P4 is designed to support ultra-low-power applications which may occasionally require high computing, by integrating a dual-core RISC-V CPU running up to 400MHz, single-precision FPU and AI extensions, and an LP-Core which can run up to 40MHz. This provides all the necessary computational resources while allowing the HP cores to be kept down for most of the time, thus saving power.

The ESP32-P4 HP core system features 768KB of on-chip SRAM, which can be used as a cache when external PSRAM is available, as well as 8KB of zero-wait TCM RAM for fast data buffers or time-critical sections of code. This powerful memory system, combined with support for external PSRAM and Flash, ensures that memory-access latency and available memory size are not restricted.

ESP32-P4 has strong security at its core, with features such as Secure Boot, Flash Encryption, cryptographic accelerators, TRNG, and more providing all the necessary components for keeping the device secure and trusted. The Digital Signature Peripheral and a dedicated Key Management Unit ensure that private keys are generated on the SoC itself and are not exposed in plain text or vulnerable to physical attacks. Additionally, the SoC supports hardware access protection, allowing for Access Permission Management and Privilege Separation.

ESP32-P4 is the ideal SoC for any HMI-based application, as it includes support for MIPI-CSI with integrated ISP and MIPI-DSI, allowing for the integration of a high-resolution camera and a display interface. Additionally, it has a parallel display and camera interface, capacitive touch inputs, and speech recognition features for enhanced compatibility. In addition, ESP32-P4 incorporates specialized hardware accelerators for multiple media-encoding and compression protocols, such as H.264 encoding, for image and video streaming. It also features a built-in Pixel Processing Accelerator (PPA) that is ideal for GUI development.

ESP32-P4 not only has more than 50 programmable GPIOs, which is more than those of any other Espressif SoC to date, but also supports all the commonly used peripherals, such as SPI, I2S, I2C, LED PWM, MCPWM, RMT, ADC, DAC, UART, and TWAITM, as well as USB OTG 2.0 HS, Ethernet, and SDIO Host 3.0 for high-speed connectivity. ESP32-P4 can be paired as a wireless companion chip for any of the ESP32-C/S/H series over SPI/SDIO/UART, via the ESP-Hosted or ESP-AT solutions if the application requires wireless connectivity. Additionally, ESP32-P4 can stand as a Host MCU for other connectivity solutions, such as ACK, AWS IoT ExpressLink, etc.

For more information about the ESP32-P4, visit the announcement page.

OKdo and RS Launch Next-Generation ROCK 5A Board

OKdo, the leading provider of next-generation computing solutions, announces the launch of in partnership with Radxa. The ROCK 5A is a powerful, versatile and easy-to-use single board computer (SBC) that is designed to meet the demands of engineers, developers, educators, inventors and entrepreneurs. Starting February 1st, OKdo will offer discounted pricing for the newly announced series of boards; limited to the first 5,000.

The ROCK 5A will be offered at:

  • 4GB – market price $99, special introductory price $74
  • 8GB – market price $119, special introductory price $94
  • 16GB – market price $159, special introductory price $134

The ROCK 5A takes the highly sought-after key functionality of its powerful predecessor, the ROCK 5B, and fits it on the industry’s common SBC form factor. Some of its key features include 8k output, an octa-core processor and an AI accelerator. Additionally, it is the only SBC on the market in this form factor that comes with 16GB of memory.

The ROCK 5A will run Linux v5.10 Kernel, supporting Ubuntu, Debian, openFyde OS, Armbian, RebornOS, Android 12 and later. Customers can download the software from OKdo.com and find additional community support by leveraging the 1.5 million members of the DesignSpark community.

Operating at a professional grade, the combination of such power, versatility and support makes the ROCK 5A an excellent option for embedded processes spanning EV charging, automatic manufacturing, digital signage, agricultural tech and more.

“Since the launch of OKdo’s family of ROCK boards, the market has been responding very well in a variety of verticals including consumer, industrial and agriculture segments. I am thrilled to add yet another member to the ROCK family and anticipate fast adoption with new and existing customers. The ROCK 5A is the perfect blend of power, versatility and affordability,” said Richard Curtin, Co-Founder and CTO of OKdo.

“We are excited to introduce the next generation of low-cost, high-performance and compact credit card-sized ROCK 5A single board computer in global collaboration with OKdo. We believe the ROCK 5A already has entry level desktop PC performance without worrying about power consumption and heating issues. We can’t wait to see the community and developers make amazing applications based on the ROCK 5A SBC,” said Tom Tang, President of VamRS Technology.

For those eager to get their hands on a next-generation SBC, the ROCK 5B is in stock at OKdo and RS for immediate shipment. Sharing key features such as the 8k output, octa-core processor, power on/off and built-in Real-Time-Clock, the 5B levels up the power with the RK3588 Rockchip.

Bringing the ROCK series to market provides a new and reliable solution to an industry still in the grips of a global chip shortage and supply chain pressures. OKdo recognizes that the time for significant change is now and that the enablement of widespread IoT adoption will be instrumental in a more sustainable future. The ROCK SBCs and CMs are another timely arm of an ecosystem of services OKdo has developed to revolutionize product development and operational excellence. For more information on ROCK 5A, please visit the OKdo website.

Launching NVIDIA Jetson Orin Nano and Orin NX edge embedded systems from Aetina

At CES 2023, Aetina announced a selection of embedded devices powered by NVIDIA Jetson Orin NX and Orin Nano. Their focus will be on the AIE-KO21/31 and AIE-KN31/41, two edge devices made for general purposes and beginner-friendly AI processing on the periphery. With mass production set to begin in Q4 2022 and Q1 2023, respectively, NVIDIA released the NVIDIA Jetson Orin NX SO-DIMM modules in March 2022 and the cheaper pin-compatible Jetson Orin Nano modules in September. Seeed Studio’s reComputer J4012 was the first NVIDIA Jetson Orin NX edge AI computer, but it turns out that Aetina also launched Jetson Orin NX/Orin Nano industrial embedded computers around the same time.

There are a lot of details on the Aetina AIE-KO21, AIE-KO31, AIE-KN31, and AIE-KN41. System-on-module in AIE-KO21 is NVIDIA Jetson Orin Nano 4 GB with up to 20 TOPS of AI performance; in AIE-KO31, it’s NVIDIA Jetson Orin Nano 8 GB with up to 40 TOPS of AI performance; in AIE-KN31, it’s NVIDIA Jetson Orin NX 8 GB with up to 70 TOPS of AI performance; and in AIE-KN41, it’s NVIDIA Jetson Or It contains a single HDMI display Type-A port and an NVMe PCIe SSD with around 128 GB of storage space. Line-out/Line-in/Mic is used for audio (optional with Daughter Board). Two external antennas and a Gigabit Ethernet RJ45 connection are supplied for wired networking.

WiFi and Bluetooth can be added through an M.2 module (See Extension). A USB Type-C (OTG) port and two USB 3.2 Gen2 Type-A ports are also included. Both the RS232 and CAN Bus serial ports have DB9 connectors, and the expansion port is isolated. In addition to the D-Sub connector with 5x GPIO, 1x I2C, 1x UART, and 1x UART for debug only (UART3), there is also a 1x M.2 E-Key 2230 (PCIe/USB2) socket for wireless connectivity and a 1x M.2 M-Key (NVMe) 2242 socket for storage. Optional functions such as a power, recovery, and reset switch are included. 12 V to 24 V DC is supplied through a 2-pin terminal block, and the usable operating temperature range is -25 °C to +70 °C, with a storage temperature range of -25°C to +85°C. The temperature is 40 degrees Celsius with a non-condensing relative humidity of 95%. Class A CE/FCC and IEC 62368 approval has been granted.

Wall mounting is possible with this enclosure, and a DIN rail mounting kit is available from Aetina if needed. It appears that the four variants of this embedded system are all the same, with only the NVIDIA Jetson Orin NX and Orin Nano modules changing their appearance. Not only does the business claim that its AIE platforms enable its EdgeEye platform for real-time remote monitoring of edge devices, but they also leave no doubt that the embedded computer will run Ubuntu 20.04-based NVIDIA JetPack.

According to Aetina, Jetson Orin NX-based systems will be made accessible in February 2023, while Jetson Orin Nano-embedded computers will follow in March. Additional information, including specifications, is available on the product website, and the press release also discusses larger Jetson Orin Nano and Orin NX systems with an additional M.2 B-Key for expanding LTE/5G functionality.

Learn how! Robotic ultraArm P340 by Arduino sketches, engraves, and grabs

The Elephant Robotics ultraArm P340 is a robot arm that has a working radius of 340 mm and is controlled by an Arduino-compatible ATMega2560 control board. The arm of this robot may be outfitted with a variety of accessories that can be used for painting, laser engraving, or grabbing objects.

The ultraArm P340 robotic arm, which is more affordable, operates somewhat differently than the myCobot 280 Pi robotic arm, which has a built-in Raspberry Pi 4 SBC. The ultraArm P340 only contains the electronics for controlling the servos and attachments, and it must be connected to a host computer running Windows or a Raspberry Pi over USB in order to function properly.

There are several different specs included in the ultraArm P340. The control board uses a Microchip ATMega2560 8-bit AVR microcontroller operating at 16 MHz. It includes 256 KB of flash memory, 4 KB of EEPROM, and 8 KB of SRAM. The degree of freedom ranges from around three to four axes, depending on the accessories. It is able to work within a radius of around 340 mm. The high-performance stepper motor allows for a weight of up to 650 grams, and the position accuracy is up to 0.1 millimeters. The maximum speed is up to 100 millimeters per second. It features two communication interfaces for RS485 and a USB serial interface. Additionally, it has attachment interfaces for a PWM laser connection, a gripper connector, and a switch connector for the suction pump. A burn button and the ability to flash the firmware are additional features.


With an input power source of 100-240 V at 50-60 Hz, the output voltage can be set to 12 V, 8.4 V, or 5 V. You may get an idea of its intricate dimensions from the image that is located above and it weighs close to 2.9 kilograms. SLA material composes its construction, which is an aluminum alloy.

The ultraArm P340 is available in a variety of configurations, including a standard kit that includes all of the fundamental accessories, a drawing kit that includes a pen holder, a laser engraving kit, and several vision kits that include a camera, clamping claw or suction pump, conveyor belt, etc… It is also feasible to purchase the extension kits separately and switch between them as you see appropriate. You can do this whenever it suits your needs.

The robot is dependent on a number of different programs and pieces of software.

  • To upgrade the firmware and gain access to user manuals, tutorials, and so on, download the myStudio software (it works on Windows, macOS, and Linux).
  • The Luban G-Code generator can be used for various artistic endeavors, including painting and laser etching.

  • In order to create Python code, myBlockly provides a visual interface for dragging and dropping blocks.
  • Python scripts may be used for direct robot control.
  • Ubuntu 20.04 appears to be hosting ROS2, the robot operating system.

The documentation website has further information and how-to-get-started guides.

The ultraArm P340 robotic arm from Elephant Robotics is now available for pre-order, with a standard kit costing $649 and shipments beginning on February 1

[Update: code ULTRA20 should reduce the price by 20%].

MyCobot 280 Pi, 6 degrees of freedom (DoF) robotic arm powered by Raspberry Pi, retails for $799 as an example product.

433 MHz is alive! Communicating with 433 MHz sensors using an ESP32 board and LoRa module.

While Matter, Bluetooth, Zigbee, and Zwave may get all the press these days, 433 MHz was the simple yet reliable technology that our sensors and devices relied on before they became popular. Given the wide range of applications for the 433 MHz, generalizations regarding the technology are likely to be oversimplified. Numerous household and automotive components employ it, including outdoor temperature and security sensors and tire pressure monitors. The various radio frequencies (315 MHz, 350 MHz, 433 MHz, 868 MHz, and 915 MHz ) and signal modulations (OOK, ASK, and FSK) encountered behind the term 433 MHz can be discussed at length. However, as 433 MHz is used by the vast majority of manufactured and commercially available sensors, they will continue to find their applications.

A 433 MHz LoRa transceiver on an ESP32 board can be used for home automation, which was demonstrated by Florian Robert of the OpenMQTTGateway project. He also discusses how to connect to 433 MHz sensors. In 2017, he published an article that compared various 433 MHz gateways for decoding 433 MHz sensors; now, he provides an update. In particular, the substantial development made by NortherMan54, who programmed an Arduino library to enable a 433MHz LoRa Transceiver-based ESP32 board to function with standard 433 MHz sensors available on the market. This provides consumers with a ready-made, off-the-shelf option for receiving these signals without breaking out the soldering iron.

LILYGO LoRa32 V2.1 and HELTEC LORa32 v2 can now be used to read OOK (On-off keying) sensors. Boards like these make it simple to install a receiver for 433 MHz sensor signals; they are based on the Espressif ESP32 processor, which many are already familiar with. They also include a Semtech SX127x Transceiver and an SSD1306 OLED display.

What really sets his work apart, though, is his ability to use a little board like this in conjunction with the RTL 433 and the RadioLib project to receive signals from a huge variety of OOK sensors. A computer or Raspberry Pi with an RTL SDR (Software defined Radio) attached would be a common setup for making use of the RTL_433 library. This method eliminates the need for an expensive USB SDR dongle while providing a low-cost, compact microcontroller-based alternative for creating a 433-to-MQTT gateway. This allows for a wide variety of sensors, such as weather stations, doors, PIR, TPMS, temperature, and BBQ sensors, to be read by a single device using the more common 433 MHz frequency.

The OpenMQTTGateway project incorporates this approach, which is necessary once the data from the sensors has been decoded and before it can be shown. Once configuring the gateway using the web portal, the data will be visible in the MQTT broker after one will web upload the application to the ESP32 board using either heltec-rtl_433 or lilygo-rtl_433. Processing them is now possible with Node-Red, Home Assistant, OpenHAB, Domoticz, and any other IoT platform that supports MQTT.

Obviously, this isn’t enough to process and show the data, so they added support for Auto-Discovery so that you do not need to manually configure an OpenHAB or Home Assistant if you don’t want to. That’s right, OpenMQTTGateway will now automatically add devices and detail their settings.

This facilitates the incorporation of the sensors into existing dashboards. Finally, functionality for showing the sensor’s data on the onboard display has been provided by NortherMan54 so you can have a fast glance at the data.

Debuting TinyML Seeed Studio Grove Vision AI Module Support in SDK, Studio By Edge Impulse

Edge Impulse, a company that specializes in machine learning, has recently made an announcement regarding a partnership with Seeed Studio. The purpose of this partnership is to add support for the Grove Vision AI Module to Edge Impulse Studio as well as the Edge Impulse software development kit (SDK), which is now available.

“Edge Impulse and Seeed have partnered together to support Seeed’s Grove Vision AI Module with Edge Impulse,” the reason why, as Joshua Buck of Edge Impulse explains it, “enabling users to now acquire, develop, and deploy vision-based ML applications with Edge Impulse Studio and Edge Impulse SDK.”

The Omnivision OV2640 camera sensor has a 1600 x 1200 UXGA resolution, while the Grove Vision AI Module is based around a low-power Himax HX6537-A processor that is running at 400MHz. The Grove Vision AI Module was introduced back in August of the previous year. The device, which is about the size of your thumb, was made for tinyML on-device computer vision processing and came pre-loaded with three image classification algorithms.

The firms have confirmed that it is now possible to enhance those three algorithms by using the Edge Impulse Software Development Kit (SDK) or the user-friendly Edge Impulse Studio. Once it has been linked to Edge Impulse Studio via USB, the Grove Vision AI Module will appear as a compatible device in the software. This will make it possible to take images for training purposes and then deploy a trained model directly to the device.

The Grove – Vision AI Module is a board around the size of a thumb that is based on the Himax HX6537-A processor. It comes with a microphone, 2-Megapixel OV2640 camera, 3-axis accelerometer, and 3-axis gyroscope. It allows individualized models in addition to providing storage with 32 MB of SPI memory, machine learning algorithms for face recognition, and people detection that are pre-installed, and it comes standard. Because it is compatible with the XIAO ecosystem as well as Arduino, it is an excellent choice for beginning work on AI-powered camera projects.

It is fully supported by Edge Impulse, which means that you will be able to sample raw data from the camera, build models, and deploy trained machine-learning models to the module directly from the studio without having to do any programming. Seeed Studio Bazaar is the exclusive location at which the Grove – Vision AI Module can be procured for financial transactions.

On the Edge Impulse website, you can find step-by-step instructions for getting started with Edge Impulse on the Seeed Grove Vision AI Module.

TOP PCB Companies