MaixCAM2 -- Rapid Deployment of AI Vision and Audio Applications

Introduction to MaixCAM2

MaixCAM2 is a hardware product designed for better implementation of AI vision, audio, and AIoT applications. It provides a platform for rapid prototype verification and mass production.

Key Features:

  1. Outstanding hardware performance: Dual-core A53 + 12.8Tops@INT4 / 3.2Tops@INT8 + 4GB LPDDR4 + multiple hardware codecs. At 640x640 resolution, YOLO11n reaches up to 113FPS, and YOLO11s up to 62FPS.
    Here are some SOC run YOLO11n's benchmark:
  2. Integrated hardware package: Supports up to 4K 1/1.8" camera, 640x480 high-definition touchscreen, dual microphones, WiFi6 + BLE5.4, and more. No complex hardware adaptation required, ready to use out of the box.
  3. Various hardware form factors: Versions available with enclosures and different accessory configurations, as well as a core board.
  4. Offline AI large model support: In addition to convolutional models, supports Transformer models, with plug-and-play LLM / VLM / ASR / TTS.
  5. Comprehensive and easy-to-use software ecosystem: Provides well-polished MaixPy (Python) + MaixCDK (C++) SDKs, extensive documentation, professional IDE, and cloud platform support.

Based on MaixCAM2, you can create:

  • DIY Smart Camera: Equipped with 4K large-format 1/1.8" sensor, supporting JPEG / RAW / H.264 / H.265 encoding, easy-to-use software, and NPU-accelerated AI features, making DIY smart cameras simple.
  • Robots: Dual mics + HD camera + HD touchscreen + AI acceleration + rich peripherals. Supports convolutional models and Transformer models with plug-and-play YOLO / LLM / VLM / ASR / TTS. Suitable for everything from toys to professional intelligent robots.
  • Production Line Quality Inspection Assistant: Traditional algorithms (OpenMV / OpenCV) plus hardware-accelerated AI recognition easily meet the demands of high-precision, real-time, and low-cost production line applications.
  • Competition Weapon: A strong tool for competitions, offering high performance, compact size, and easy-to-use Python (MaixPy) and C++ (MaixCDK) SDKs. Enables lightning-fast development of competitive projects. Many students have already won top awards using MaixCAM.
  • Educational Aid: For university research, teaching, STEM education, spreading cutting-edge knowledge, and inspiring future technologies.
  • And more: Waiting to be discovered by you!

Feature Showcase

For more feature introductions, please refer to: MaixPy 主页

MaixCAM2 Hardware Specifications

Bold items are upgrades compared to MaixCAM / MaixCAM-Pro (first generation).

Component Description
CPU (Big Cores) 1.2GHz A53 x2, runs Linux (Ubuntu)
CPU (Small Core) RISC-V 32bit E907, runs RTT
NPU 12.8Tops@INT4 / 3.2TOPS@INT8, supports convolution and Transformer models such as YOLO/LLM/VLM, YOLO11n 640x640 reaches up to 113FPS
Memory 1GB / 4GB LPDDR4 options
Storage Onboard 32GB eMMC, onboard TF card slot
Camera Supports up to 8MP (4K)@30fps, 4-lane MIPI CSI input, 22-pin interface, supports split dual CSI
Display 2.4-inch HD IPS capacitive touchscreen, resolution 640x480 (4-lane MIPI DSI output, standard 31-pin interface, 6-pin capacitive touch), supports up to 1080p@60fps output
Audio Output Onboard PA amplifier + 1W speaker
Audio Input Onboard analog silicon mics x2, direct audio capture
Networking Onboard WiFi6 + BLE5.4 module, 6-pin FPC Ethernet interface (with external FPC-to-RJ45 module)
USB Type-C USB2.0, supports both Device and Host modes, supports USB cameras
IO Interface 2.54mm PMOD interface, provides 20 IOs + Vsys/3.3V/GND + 1.25mm 6-pin expansion interface, core board offers more IOs
Buttons 1 x Power switch + 1 x Func (function) button
LEDs Power indicator + User LED + Illumination LED
Codec H.264 / H.265 / MJPEG hardware codec, supports 4K@30fps encoding, 1080p@60fps decoding
Peripherals I2C / SPI / UART / ADC / PWM / WDT and other common peripherals
Power Supports Li-ion battery charge/discharge management, with battery-equipped version available
Enclosure Protective case with 1/4-inch standard tripod mount
IMU Onboard six-axis IMU (3-axis accelerometer + 3-axis gyroscope)
RTC Onboard BM8563EMA RTC chip + rechargeable coin battery, maintains time during power-off
Core Board Gold-finger core board containing only chip core circuit + DDR, for easy custom hardware design

MaixCAM2 Software Ecosystem

We don’t just provide hardware — MaixCAM2 comes with a complete software ecosystem, including:

Name Description Image/Video
MaixPy Python SDK with rich and easy-to-use APIs, optimized for MaixCAM, supports hardware acceleration, and comes with comprehensive documentation and tutorials 1. MaixPy Homepage
2. MaixPy Source Code
MaixVision AI vision IDE for coding, code execution, real-time image preview, even graphical programming. Greatly lowers the difficulty and barriers to setup. MaixVision
MaixHub Online AI model training platform. No AI knowledge or expensive hardware required — train models with one click and deploy them to MaixCAM instantly MaixVision
MaixCDK C++ SDK equivalent of MaixPy. Developers familiar with C/C++ can get started immediately See MaixCDK Homepage
App Store Various apps and tools available for download without development. Developers can also upload and share apps See MaixHub App Store
Sharing Hub A community for developers to share experiences and projects See MaixHub Sharing Hub

Resources

MaixCAM Exclusive Resources (Provided by Sipeed)

Chip Resources

MaixCAM2 is based on the AX630C chip from Axera, so their resources can also be referenced.

Accessories

Community

Purchase