¶ BPI-OM7 3D Camera
¶ Key Features
The Banana Pi BPI-OM7 is an integrated AI 3D vision computing platform consisting of the Banana Pi BPI-M7 development board and an ORBBEC Gemini 2 3D camera.
Based on the high-performance RK3588 SoC, this platform combines active binocular infrared depth sensing technology, targeting applications in 3D vision, robotics, edge AI, and spatial perception.
The BPI-OM7 provides a complete hardware foundation, from RGB/Depth/IMU data acquisition to local AI inference and edge computing. It supports multiple mainstream operating systems and AI frameworks, making it suitable for development, testing, and product deployment.
| Rockchip RK3588 design Banana Pi BPI-M7 AI SBC design |
| ORBBEC Gemini 2 3D Camera |
¶ Orbbec Gemini 2
The Orbbec Gemini 2 is a binocular structured light 3D camera powered by the Orbbec MX6600 depth engine chip. It features three depth modes, providing high-quality depth data for various applications. It offers a wide field of view, with a depth image measurement range of 0.15-10m, and integrates auxiliary point ranging functionality, enabling zero-blind-zone depth measurement up to 10m.
The Gemini 2 is small yet powerful, featuring a built-in six-axis IMU and integrated hardware D2C functionality (depth and RGB image pixel-by-pixel alignment), significantly reducing the computational requirements of the host computer. It also supports flexible multi-camera synchronization, providing a wider field of view and more comprehensive data, easily adaptable to various sensing and measurement solutions.
¶ Key Specifications
-
Main controller uses Rockchip RK3588 4x Cortex-A76 @ 2.4GHz + 4x Cortex-A55 @ 1.8GHz, 8nm
-
8GB LPDDR4 memory
-
64GB eMMC storage
-
Hardware D2C: Significantly reduces the computing power requirements of the host computer
-
High precision and large FoV, depth diagonal FoV greater than 100°
-
Zero blind zone, providing depth measurement values at extremely close distances of 0-0.4m
-
Multi-machine synchronization, enabling multi-machine combined linkage
-
Compact size, integrating a high-performance 6-axis IMU
¶ Application scenarios
-
Automated sorting
-
Moving environment perception
-
3D human/object reconstruction
-
Interactive fitness
-
Developer development
-
Intelligent spatial perception and people flow analysis
-
AI and computer vision teaching and research experimental platform
¶ Getting Start
¶ Hardware
¶ BPI-OM7 3D Camera Photo
¶ Hardware Specifications
¶ BPI-M7
| HardWare Specification of Banana Pi BPI-M7 | |
|---|---|
SOC |
RockChip RK3588 |
CPU |
RK3588 Quad-Core Cortex-A76@ 2.4GHz+Quad-CoreCortex-A55@ 1.8GHz，8nm process |
GPU |
GPU Mali-G610 MP4 (4x256KB L2 Cache) Supports OpenGL ES3.2/OpenCL2.2/Vulkan1.1 |
NPU |
6 TOPS@INT8(3 NPU core) |
Supports INT4/INT8/INT16 mixed computing |
|
Supports frameworks like TensorFlow, MXNet, PyTorch, Caffe, Tflite, Onnx NN, Android NN, etc |
|
VPU/Codec |
Hardware Decode: 8K@60fps H.265/VP9/AVS2、 8K@30fps H.264 AVC/MVC、 4K@60fps AV1、1080P@60fps MPEG-2/-1/VC-1/VP8 |
Hardware Encode:8K@30fps H.265 / H.264 |
|
ISP |
Integrated 48MP ISP with HDR&3DNR |
RAM |
8GB/16GB/32GB（max 32GB ）64bit LPDDR4/LPDDR4x，default 8GB LPDDR4x |
Flash |
32GB/64GB/128GB eMMC，default 64GB eMMC,Supports MicroSD card expansion |
PCIe |
1x M.2 Key M (PCIe 3.0 4-lanes). Expandable SSD support for 2280/2260/2242/2230 M.2 SSD cards, default 2280 |
Networking |
2x 2.5G Ethernet |
Onboard IEEE 802.11a/b/g/n/ac/ax WIFI6 and BT5 (AP6275P) |
|
Video Out |
1x HDMI 2.1, supports 8K@60fps |
1x MIPI DSI up to 4K@60fps |
|
1x DP 1.4 up to 8K@30fps |
|
Video In |
2x 4-lane MIPI CSI, up to 2.5Gbps per lane |
Audio |
1x HDMI audio out |
1x HP audio out |
|
1x Type-C 3.1 (DP1.4) audio out |
|
USB |
1x USB 3.0 (USB 3.1 Gen 1), equivalent to USB 3.2 Gen 1/USB 3.0, up to 5Gbps |
1x USB Type-C 3.1（DP1.4/OTG） |
|
1x USB 2.0 High（480Mbps）/Full（12Mbps）/Low-Speed（1.5Mbps） modes |
|
40-pin |
Fully compatible with Raspberry Pi 40-pin header for connecting abundant add-on modules |
Supports UART/SPI/I2C/I2S/PWM/ADC/5V Power/3.3V Power |
|
Other |
1x 5V fan interface |
1x battery connector for low power RTC chip HYM8563TS |
|
2x LEDs - blue LED blinks on system start, red LED user controllable |
|
Power Input |
USB Type-C PD 2.0, 9V/2A, 12V/2A, 15V/2A |
Buttons |
1x PWRON button for sleep/wake,1x Reset button for reboot,1x Maskrom button for maskrom burn-in mode |
OS Support |
Official: ：Android 12.0，Debian11，Buildroot |
3rd Party：Armbian，Ubuntu 20.04，Ubuntu22.04，Kylin OS |
|
Dimensions |
92 mm x 62mm |
Operating temperature |
0℃ ~ 80℃ |
¶ ORBBEC Gemini 2
| Camera Specifications | |
|---|---|
Depth Technology |
Active Stereo IR |
Wavelength |
850nm |
Depth Range |
*0.15-10m |
Depth Resolution/FPS |
Up to 1280x80@30fps |
Depth FOV |
H91°V66° |
RGB Resolution/FPS |
Up to 1920x1080@30fps |
RGB FOV |
H86°V55° |
Processng |
Orbbrc ASIC |
IMU |
Supported |
*Precision:≤2%@2m |
|
¶ Development
¶ Source code
| BPI-M7 github source code： https://github.com/ArmSoM/armsom-build |
| BPI-M7 kernel code: https://github.com/ArmSoM/linux-rockchip |
| BPI-M7 uboot code ： https://github.com/ArmSoM/u-boot |
| openwrt(istoreos) ： https://github.com/istoreos/istoreos |
| RKNN-LLM Code repository: https://github.com/ArmSoM/rknn-llm |
¶ Resources
| BPI-M7 Component diagram |
Baidu Cloud: https://pan.baidu.com/s/1Klh1xt_2Qkd8ZKZ6EIbEbg?pwd=8888 （pincode:8888)
| BPI-M7 2D CAD |
Baidu Cloud: https://pan.baidu.com/s/1spHZGhDMC4TW9G9HTei7rA?pwd=8888 (pincode:8888)
| BPI-M7 Schematic |
Baidu Cloud: https://pan.baidu.com/s/17_2BTwBnIsmeCV5V7xNXRA?pwd=8888 （pincode:8888)
| Rockchip RK3588 datasheet : |
Baidu cloud: https://pan.baidu.com/s/1GeskKfLFwjgmz0pgt7sICg?pwd=8888 (pincode:8888)
| BPI-M7 GPU and CPU performance test: https://www.youtube.com/watch?v=C4ofIZLixpM |
¶ System Image
¶ Easy to buy
| BANANAPI Official Shop: |
| Taobao Shop: |
| Bipai Aliexpress shop: |
| SINOVOIP Aliexpress shop: |
| OEM&ODM, please contact: [email protected] |