Edge AI video processing system combines SocioNext SC2000 4K camera SoC with Hailo-8 AI accelerator

EdgeTuring AI processing solution

We’ve previously written about Hailo-8 AI accelerator delivering up to 26 TOPS at low power (3 TOPS per watt), and found in AI edge systems such as Foxconn BOXiedge server powered by a Socionext SynQuacer SC2A11 24x Cortex-A53 cores SoC and capable of analyzing up to 20 streaming camera feeds in real-time for video analytics. Leopard Imaging has now created a much more compact Edge AI processing solution with EdgeTuring based on Hailo-8 M.2 card and Socionext SC2000 4K camera SoC, and designed for low-power video analytics. EdgeTuring features and specifications: Camera SoC / Image Processor – Socionext SC2000 quad-core Cortex-A7 processor @ 650MHz with 4Kp30 H.265/H.264 encoder, up to 1.2 Gigapixel per second processing, LPDDR-3  memory (PoP) AI Module – Hailo-8 M.2 Module for up to 26 TOPS  Camera Dual camera but only one sensor works for AI Sensors – 2x Sony IMX477 CMOS Image Sensor, diagonal 7.857 mm […]

ANNKE CZ400 AI security camera review – Part 1: Specs, unboxing and teardown

Annke CZ400 AI security camera review

We’re starting to see more and more security cameras with built-in AI features such as Reolink RLC-810A 4K camera with people & vehicle detection or the lower cost Vacom Cam Full HD camera with person detection only, which I have not reviewed yet due to some technical issues. I would not buy a camera without AI now as such features greatly reduce false positives. Another model has now come my way with ANNKE CZ400 (aka I91BK) 2K security camera that comes with even more advanced AI-accelerated computer vision features with not only face detection, but also intrusion detection, line crossing detection, region entrance detection, region exiting detection, object removal detection, unattended baggage detection, and audio exception detection. In this first part of the review, we’ll check the specifications, the content of the package with an unboxing, as well as a partial teardown that is needed to install a microSD card. […]

tCam-Mini IR thermal camera board ships with ESP32 module

tCam mini

While there are plenty of ESP32 camera boards, it’s much harder to find off-the-shelf solutions with ESP32 wireless SoC and an IR thermal camera. That’s likely why Dan Julio decided to design tCam-Mini board combining an ESP32 module with a Flir Lepton 3.5 sensor with 160×120 resolution to capture radiometric data for thermographic analysis. tCam-Mini board hardware specifications: Wireless module – Espressif ESP32-WROVER-E module with dual-core ESP32-D0WD0V3 WiSoC, 8 MB PSRAM, 8 MB Flash, and PCB antenna; Camera – Flir Lepton 3.5: 160×120 pixel radiometric LWIR camera with shutter operating in either Radiometric/TLinear (each pixel contains temperature data) or AGC modes (no temperature data in each pixel but better images). Expansion – 4-pin I2C header Debugging – Micro USB port with CP2102N-A02 USB to UART bridge for ESP32 boot loader control Misc – Dual color (Red/Green) status LED, Factory/Wifi Reset button Power supply – 5V input via micro USB port […]

StereoPi v2 stereoscopic camera is powered by Raspberry Pi CM4 (Crowdfunding)

StereoPi v2 Raspberry Pi CM4 camera

StereoPi stereoscopic camera based on Raspberry Pi Compute Module 3 was introduced in late 2019 on Crowd Supply.  The camera can record 3D video, create 3D depth maps with OpenCV, and benefits from the Raspberry Pi software ecosystem. The developers are now back with an upgraded model. StereoPi v2 comes with many of the same features, but as it is based on Raspberry Pi CM4 (Compute Module 4) it offers better performance, Gigabit Ethernet, Wifi & Bluetooth connectivity out of the box, while other features like PoE, TFT screen, shot button, etc.. are optional. StereoPi v2 specifications: Supported SoM – Raspberry Pi CM4 or CM4Lite modules Storage – MicroSD card socket Video Output – Micro HDMI port Camera I/F – 2x MIPI CSI camera connector plus “hackable camera lines” Networking – Gigabit Ethernet RJ45 port, plus optional WiFi 5 and Bluetooth 5.0 on Raspberry Pi CM4 module USB – 2x […]

Vacos Cam AI Security Camera Review – Part 1: Specifications, Unboxing and Teardown

Vacos Cam teardown

As we’ve seen in our Reolink RLC-810A review, AI security cameras greatly reduce the number of false alerts generated by motion sensors, and the Reolink 4K security camera we tested was capable of people and vehicle detection. The Reolink model does not support WiFi, so I had to run a long Ethernet cable, plus a power extension to connect a 12V power adapter since I don’t own a PoE injector. But I’ve been given the opportunity to review another AI security camera, namely Vacos Cam, that also supports human detection, but relies on WiFi connectivity and can be powered by a solar panel removing the need for potentially long cable. As a bonus, it comes with eMMC for storage, so there’s no need for a MicroSD card. In the first part of the review, I’ll list the specifications, unbox the power panel and camera, and tear down Vacos Cam to […]

Jetson AGX Xavier based Edge AI Platform supports 8 GMSL automotive cameras and 10Gbps Ethernet

Jetson Xavier AI Edge computer with 8 GSML cameras

Neousys Technology, a Taiwan-based provider of rugged embedded systems, has introduced the NRU-110V Edge AI inference platform powered by an  NVIDIA Jetson AGX Xavier SoM. It comes with eight FAKRA Z connectors to support up to eight GMSL automotive cameras, as well as one 10GbE (10 Gigabit Ethernet) port for real-time video data transfer. The system specifically supports GMSL cameras with 3-exposure HDR, LED flicker mitigation, good light-sensitivity, and IP67 waterproof characteristics,  and also comes with a synchronize mechanism to simultaneously capture high-quality images from 8 GMSL cameras while accepting GPS PPS signal to align image data with other sensors. That makes it suitable for various applications ranging from autonomous driving, mobile robots, precision agriculture, intelligent V2X, to teleoperation. Neousys NRU-110V specifications: System-on-Module – NVIDIA Jetson AGX Xavier SoM with SoC NVIDIA 512-core Volta GPU 8-core Arm Carmel CPU 2x NVDLA Engines 7-Way VLIW Vision Processor Memory – 32GB LPDDR4x […]

Person Detection on Raspberry Pi Pico with ArduCAM and TensorFlow Lite

ArduCAM with Raspberry Pi Pico

ArduCAM is popular for camera-based applications with various boards ranging from Arduino to Raspberry Pi. We also saw the company’s tiny coin-sized Raspberry Pi compatible module 5 years ago. Now, it also supports the newly launched Raspberry Pi Pico for real-time video applications. Raspberry Pi Pico is compatible with the ArduCAM Mini 2MP Plus camera featuring an OV2640 2MP CMOS image sensor that supports automatic image control functions including Automatic Exposure Control (AEC) and Automatic Gain Control(AGC). The camera also comes with an onboard JPEG encoder for image compression. The company has provided a Github repository with two demo applications: a video streaming application and an example for basic person detection with the probability percentage of detection. There is also an option of directly using the UF2 files for flashing with Raspberry Pi Pico, if you don’t want to build the demo from the source code yourself. The application runs […]

Ingenic T40 4K Video & AI Vision Processor mixes MIPS & RISC-V cores with AI accelerator

Ingenic T40 MIPS RISC-V AI

Ingenic T31 MIPS & RISC-V AI video processor was introduced last year with the MIPS core including SIMD128 Vector instructions for deep learning. The Chinese company has now introduced the new Ingenic T40 processor still with MIPS and RISC-V cores, but also a dedicated 8 TOPS AI engine/CNN accelerator. The new processor is especially suited to smart AI vision application thanks to support for 4K cameras, and 4K MJPEG/H.264/H.265 hardware video encoding, complemented by the 8 TOPS AI engine for computer vision workloads such as people detection, face recognition, object detection, and so on. Ingenic T40 key features and specifications: CPU – Dual-core MIPS XBurst2 @ 1.2 GHz with 256KB L2 Cache, SIMD512 instruction set MCU – 600MHz RISC-V coprocessor AI Engine – 8 TOPS neural network accelerator with 1MB memory pool, support for int16/int8/int4/int2 convolution width Memory – DDR2/DDR3/DDR3L up to 2GB Storage – SPI NOR flash, SPI NAND flash, […]

EDATEC Raspberry Pi 5 fanless case