An industrial edge AI box built on the BM1684 (17.6 TOPS INT8) that analyses 16 HD camera streams at once, with 3× GbE, isolated CAN, RS-485, DIO and 12–36V DC input.
Overview
The YZ-B011 turns existing IP cameras into an AI safety and operations system without sending video off-site. Its BM1684 SoC (8× Cortex-A53 at 2.3 GHz, 12 nm) delivers 17.6 TOPS INT8 and decodes up to 32 channels of 1080p@30 while running deep-learning inference on 16 HD streams simultaneously. Three Gigabit ports, isolated CAN, RS-232/RS-485 and opto-isolated digital I/O let the box trigger alarms, barriers and PLCs directly. More than 50 ready-made algorithms are available, and models from TensorFlow, PyTorch, ONNX, Caffe, MXNet and PaddlePaddle can be deployed.
Key features
- BM1684: 8× Cortex-A53 @ 2.3 GHz; 17.6 TOPS INT8 / 2.2 TFLOPS FP32
- Analyses 16 HD streams simultaneously; decodes up to 32× 1080p@30; MJPEG 1080p@480 fps
- 12/16 GB LPDDR4(X); 32/64/128 GB eMMC; M.2 SATA (2242); TF
- 3× Gigabit Ethernet; 2.4/5 GHz Wi-Fi; 4G/5G via M.2
- 2× USB 3.0, 2× USB 2.0, HDMI, audio in/out
- 4× opto-isolated DIO (2 in / 2 out), RS-232, RS-485, 2× isolated CAN
- DC 12–36V wide-voltage input; -20 to +70 °C
- 50+ algorithms: fire and smoke, intrusion, crowding, PPE (helmet, uniform), illegal parking, blocked fire lanes, e-bikes in elevators, overflowing bins and more than 40 others
Specifications
| Item | Specification |
|---|---|
| AI SoC | SOPHGO BM1684, 12 nm |
| CPU | 8× Arm Cortex-A53 @ 2.3 GHz |
| AI Performance | 17.6 TOPS INT8 (35.2 TOPS INT8 Winograd); 2.2 TFLOPS FP32 |
| Video Decode | Up to 32× 1080p@30 (H.264/H.265); MJPEG 1080p@480 fps |
| Video Encode | 1080p@50 |
| Concurrent Analytics | 16 HD streams |
| Memory | 12 / 16 GB LPDDR4(X) |
| Storage | 32 / 64 / 128 GB eMMC; 128 MB SPI flash; M.2 SATA 2242; TF |
| Ethernet | 3× Gigabit |
| Wireless | 2.4/5 GHz Wi-Fi; 4G/5G (M.2, optional) |
| USB | 2× USB 3.0, 2× USB 2.0 |
| Industrial I/O | 4× DIO (2 DI / 2 DO, opto-isolated), 1× RS-232, 1× RS-485, 2× isolated CAN |
| Display / Audio | HDMI; audio in/out |
| Frameworks | TensorFlow, Caffe, MXNet, PyTorch, ONNX, PaddlePaddle |
| OS | Ubuntu 20.04 |
| Power Input | DC 12–36V |
| Operating Temperature | -20 to +70 °C |
| Storage Temperature | -40 to +85 °C |
| Humidity | 5–95% RH |
Specifications are based on manufacturer data and may change. Items marked “on request” are configured per project; ask for the latest datasheet.
Applications
- Industrial parks and factories (PPE, intrusion, fire/smoke)
- Fuel stations (smoking, phone use, unattended nozzles)
- Battery and EV charging sheds (smoke/fire detection)
- Residential communities (e-bikes in elevators, fire lanes)
- Construction sites, mines, ports and warehouses
Related industry solutions: AI Video Analytics & Smart Safety · EV Charging Stations · Fire Safety & Building Control
Customization and ordering
The YZ-B011 can be customized for OEM projects: interfaces, enclosure or mounting, firmware/OS image and branding. Small-batch pilot orders are supported. Request a quote or sample, chat on WhatsApp (+86 135 0532 6001), or read about our OEM/ODM service.
More in this category: Edge AI Computers & Video Analytics Boxes
Frequently asked questions
How many cameras can the YZ-B011 analyse?
The YZ-B011 runs AI analytics on 16 HD streams at the same time and can decode up to 32 channels of 1080p video.
Can I deploy my own AI models?
Yes. Models from TensorFlow, PyTorch, ONNX, Caffe, MXNet and PaddlePaddle can be converted and deployed; Docker and Python development are supported.
Does video leave the site?
No. Analysis runs locally on the box, so video stays on the local network and only events or alarms are forwarded.
What power supply does it need?
It accepts a wide 12–36V DC input, suitable for industrial cabinets.