High-quality Customized OEM/ODM Paris Edge Computing Box in High-end Factory & Manufacturers-HUAXIN CHUANGHE

◆Utilizes the highly integrated CV186AH intelligent vision deep learning chip
◆Powered by a 6-core ARM A53 processor
◆1.6GHz clock speed, providing 7.2TOPS@INT8 computing power
◆Supports intelligent analysis of 8-channel HD video
◆CCC certified and meets CANS-related testing requirements
◆IP50 protection rating

The ThingSense SDK², a development tool supporting the ThingSense platform, provides a unified interface definition and development framework, shielding against differences in hardware chipsets and operating environments. This allows developers to quickly develop algorithm applications compatible with the ThingSense platform, allowing them to focus solely on core business logic.

It provides a unified model loading/calling interface, model conversion tools, and compilation and release tools. Algorithm developers only need to focus on chip-compatible model training, core business logic, and data interfaces, enabling one-click model conversion, app compilation, packaging, and image deployment. By abstracting terminal devices into standardized data source interfaces, algorithms can directly access device data and automatically convert inference results into a unified format for reporting to the ThingSense platform.

Unified Video Decoding (SRE) provides a unified decoding cache for terminal video streams, allowing multiple algorithm apps to share the data, saving video decoding resources.

Core Configuration	Chip Platform	SOPHON CV186AH
	Processor	6-core ARM Cortex A53 @ 1.6GHz
	Hashrate	7.2 TOPS@INT8 / 1.5T@FP16/BF16
	ISP	Supports time-division multiplexing for processing multiple sensor inputs and 3A (AE/AWB/AF) functions with user-adjustable 3A parameters. Supports two-frame wide dynamic range and Advanced Local Tone Mapping, fixed-pattern noise reduction, defective pixel correction, lens distortion correction, purple fringing correction, Bayer noise reduction, 3D noise reduction, detail enhancement, and sharpening enhancement.
	Memory	Default 8G LPDDR4, optional 4G / 16G
	Storage	Default 32G eMMC, optional 64G / 128G
	Expanded storage	M.2 SSD: Supports NVMe SSDs 2242, 2260, and 2280 (internal to the chassis) SATA: Supports expansion of SATA hard drives (internal to the chassis) TF: Supports TF memory cards
	Video Codec	Decoding: H.265 & H.264, 16-channel 1080P @ 30fps Encoding: H.265 & H.264, 10-channel 1080P @ 30fps
	Image codec	Decoding: 480 fps @1080P; Encoding: 480 fps @1080P
Basic parameters	power supply	DC 12VDC/3A
	Heat dissipation method	Passive cooling
	Power consumption	Typical power consumption: 12W; Peak power consumption: 18W
	Environmental parameters	Working temperature: -20℃~60℃; Working humidity: 10%~90%RH, non-condensing
	Structural dimensions	204mm × 149mm ×64mm
	Protection level	IP40
Interface parameters	Network interface	RJ45 x2: Supports 10/100/1000Mbps network access Wi-Fi: Supports expansion of Wi-Fi/Bluetooth modules via the M.2 interface 4G/5G: Supports expansion of 4G/5G modules via the M.2 B-KEY interface
	Video Output	HDMI OUT × 1 (supports 4K@60fps video output) Note: Only supports direct connection
	Audio Interface	Line_IN × 1: audio input, Line_OUT × 1: audio output (standard 3.5mm audio interface)
	Other interfaces	USB3.0 × 2 / HDMI × 1 / TF × 1 / SIM × 1 / RS-485 × 2 / RS-232 × 1 / CAN × 1 / GPIO × 2 / DI × 1 / DO × 1 / DEBUG × 1 / External power supply (5V) × 1 / External power supply (12V) × 1
	Extensible interface	M.2（4G/5G） ×1 / M.2（ SSD） ×1 / M.2（WI FI+BT） ×1 / SATA × 1
Software Configuration	System version	Ubuntu 22.04 / 20.04 optional
Software Configuration	Software Support	Supports multiple deep learning frameworks, including TensorFlow, ONNX, Caffe, TFLite, Pytorch, MxNet, PaddlePaddle, and Dark Net. Supports private deployment of large language models such as Gemma-2B, LlaMa2-7B, ChatGLM3-6B, and Qwen1.5-1.8B using the Transformer architecture. Supports Docker container management technology.

Our Services Category

Edge Computing Box HX-072AI