Equipment and materials for education, makers and small businesses.
Product imagery based on manufacturer photography.
Equipment or materials in application photos may be sold separately.
Explore computer vision with a camera that recognises objects, gestures and more. Train an object model, test it on the device and connect compatible robotics projects.
Available to order from manufacturer
Estimated delivery: Fri 6 Nov
Makeblock · Vision modules
AI Camera 2.0 brings computer vision to a small, self-contained module with a camera, touchscreen and onboard processing. Try its built-in recognition modes, inspect the result on the display and discuss what the camera has recognised. Learners can also collect examples and train their own object-recognition model, then test how lighting, distance and the background affect its predictions.
Use the camera on its supplied tripod for an introductory activity, or connect it to compatible Makeblock hardware for a project that responds to visual information. In mBlock, blocks and Python link recognition results to programmed actions. This camera-only package includes its mounting and activity accessories; an mBot2, CyberPi or other robot controller is supplied separately when the project needs one.
| Brand | Makeblock |
|---|---|
| Configuration | Camera only |
| Audio | Microphone and 1 W speaker |
| Camera | 12 MP · manual focus |
| Memory | 256 MB DDR3 |
| Display | 2.4-inch IPS capacitive touchscreen · 640 × 480 pixels |
| Storage | 32 GB TF card |
| Controls | Learn button, three-way mode switch and power switch |
| Lighting | Integrated illumination LED |
| Mounting | ¼-inch tripod mount and Makeblock mounting accessories |
| Controller | RISC-V C906 processor · 1 GHz |
| Dimensions | 70 × 54 × 21 mm |
| Programming | Block-based coding and Python in mBlock |
| AI processor | 1 TOPS neural processing unit at INT8 |
| Connectivity | Wi-Fi 6, USB-C (USB 2.0) and mBuild |
| Construction | Translucent housing |
| Motion sensors | Six-axis IMU with accelerometer and gyroscope |
| Custom training | Object recognition |
| Expansion ports | Two mBuild, one I²C, one UART and two PWM ports |
| Recognition modes | Face, colour, gesture, tag, object, path, posture, text (OCR) and voice |
| Supported USB devices | Compatible USB cameras |
| Product family | Vision modules |
| Applications | Code |
Camera only
No. This is the camera-only package with its standard cables, mounting parts, activity items and tripod. An mBot2 or another robot is sold separately.
Yes. Its display and onboard processing allow standalone recognition activities. Connect a compatible robot or controller when you want the recognition result to control another device.
Yes, for object recognition. Collect examples, train the model and test it on the camera. The other built-in recognition modes are not custom-training modes.
The camera provides I²C and serial interfaces for integration projects. These require appropriate wiring and software; they are not the same as connecting a ready-made mBuild project.