1. Introduction
MaixCAM2 is a hardware platform designed for rapid development and deployment of AI vision, AI audio, and AIoT (Artificial Intelligence of Things) applications. It enables developers to quickly prototype AI-powered products and seamlessly transition to mass production.
1.1 Key Features
High-Performance Hardware
- Dual-core Arm Cortex-A53 processor
- AI performance of 12.8 TOPS (INT4) or 3.2 TOPS (INT8)
- Up to 4 GB LPDDR4 memory
- Multiple hardware video encoding and decoding engines
- Runs AI models independently at 640 × 640 resolution
- Delivers up to 113 FPS with YOLO11n
- Achieves up to 62 FPS with YOLO11s
Integrated Hardware Platform
- Supports up to a 4K 1/1.8-inch image sensor
- 640 × 480 high-definition capacitive touchscreen
- Dual microphones
- Built-in Wi-Fi 6 and Bluetooth 5.4
- Ready to use without complex hardware integration
Multiple Hardware Options
- Available as a complete device with enclosure
- Multiple accessory configurations
- Core board version available for custom hardware development
Offline AI Large Model Support
Supports both convolutional neural networks and Transformer-based AI models, including:
- LLM (Large Language Models)
- VLM (Vision Language Models)
- ASR (Automatic Speech Recognition)
- TTS (Text-to-Speech)
Comprehensive Software Ecosystem
- MaixPy (Python SDK)
- MaixCDK (C++ SDK)
- Extensive documentation
- Professional IDE
- Cloud platform support
1.2 Applications
DIY Smart Camera
- Supports up to 4K cameras with a 1/1.8-inch sensor
- JPEG, RAW, H.264, and H.265 encoding
- AI-accelerated image processing using the onboard NPU
- Ideal for building intelligent cameras
Intelligent Robotics
- Dual microphones
- HD camera
- Capacitive touchscreen
- AI hardware acceleration
- Supports YOLO, LLM, VLM, ASR, and TTS
- Suitable for educational, consumer, and industrial robots
Industrial Quality Inspection
- Combines traditional computer vision (OpenMV/OpenCV) with AI recognition
- High accuracy and real-time performance
- Cost-effective solution for production line inspection
AI Competition Platform
- High performance in a compact form factor
- Easy-to-use Python (MaixPy) and C++ (MaixCDK) SDKs
- Ideal for robotics, AI, and embedded system competitions
- Successfully used by students to win top awards in various competitions
Education and Research
- University research
- STEM education
- AI learning and experimentation
- Embedded system development
More Possibilities
The flexible platform enables developers to create countless additional AI-powered applications.
1.3 Function Demonstration
For a complete list of software features and examples, please refer to the MaixPy documentation.
2. MaixCAM2 Hardware Specifications
| Component | Specification |
|---|---|
| Processor (SoC) | Axera AX630C |
| High-Performance CPU | Dual-core Arm Cortex-A53 @ 1.2 GHz (Linux/Ubuntu) |
| Real-Time Core | 32-bit RISC-V E907 running RT-Thread (RTT) |
| AI NPU | 12.8 TOPS (INT4) / 3.2 TOPS (INT8), supports CNN and Transformer models including YOLO, LLM, and VLM |
| AI Performance | YOLO11n up to 113 FPS at 640 × 640 resolution |
| Memory | 1 GB or 4 GB LPDDR4 |
| Storage | 32 GB onboard eMMC, microSD (TF) card slot |
| Camera Interface | Supports up to 8 MP (4K @ 30 FPS), 4-lane MIPI CSI, 22-pin connector, dual CSI support |
| Display | 2.4-inch IPS capacitive touchscreen, 640 × 480 resolution, MIPI DSI, supports external display up to 1080p @ 60 FPS |
| Audio Output | Built-in 1 W speaker with power amplifier |
| Audio Input | Dual onboard analog MEMS microphones |
| Networking | Wi-Fi 6, Bluetooth 5.4, optional Ethernet via FPC-to-RJ45 adapter |
| USB | USB Type-C (USB 2.0), supports Host and Device modes, compatible with USB cameras |
| Expansion I/O | PMOD connector, 20 GPIOs, UART, SPI, I²C, ADC, PWM, and additional expansion connector |
| Buttons | Power button and programmable Function button |
| LED Indicators | Power LED, User LED, and Illumination LED |
| Video Codec | Hardware H.264, H.265, MJPEG decoding, 4K @ 30 FPS encoding, 1080p @ 60 FPS decoding |
| Peripherals | I²C, SPI, UART, ADC, PWM, Watchdog Timer |
| Power | Integrated lithium battery charging and power management |
| Enclosure | 66.5 × 50 × 21.2 mm (excluding lens), standard 1/4-inch tripod mounting thread |
| IMU | Built-in 6-axis sensor (3-axis accelerometer + 3-axis gyroscope) |
| RTC | BM8563EMA Real-Time Clock with rechargeable backup battery |
| Core Module | Available as a standalone compute module for custom hardware designs |
3. MaixCAM2 Software Ecosystem
MaixCAM2 provides a complete software ecosystem in addition to its hardware platform, including:
- MaixPy – Python SDK for rapid AI application development.
- MaixCDK – C++ SDK for high-performance embedded applications.
- Comprehensive documentation and tutorials.
- Professional integrated development environment (IDE).
- Cloud platform integration.
- Rich AI model support, including YOLO, LLM, VLM, ASR, and TTS.
- Extensive sample applications and development resources.
- Fast deployment from prototype to production-ready AI devices.
