SSCMA Micro Core for Arduino

SSCMA Micro Core for Arduino is a library designed for Arduino, which supports a variety of AI algorithms (Classification, Detection, Pose Estimation, etc.). The library provides a simple and easy-to-use interface for developers to seamlessly integrate SSCMA-Micro Core with their Arduino projects.

Installation

1. Download the latest version of the library from our GitHub repository.
2. Add the library to your Arduino IDE by selecting Sketch > Include Library > Add .ZIP Library and choosing the downloaded file.

Usage

1. Use SenseCreaft AI Platform or other tools to flash a compatiable model to your device.

2. Include the library in your Arduino sketch:

   #include <SSCMA_Micro_Core.h>

3. Create an instance of the SSCMAMicroCore:

   SSCMAMicroCore instance;
   SSCMAMicroCore::VideoCapture capture;

4. Initialize the SSCMAMicroCore and register callback functions in setup() function:

   void setup() {
       capture.begin(SSCMAMicroCore::VideoCapture::DefaultCameraConfigXIAOS3);
       instance.begin(SSCMAMicroCore::Config::DefaultConfig);
       instance.registerPerfCallback(SSCMAMicroCore::DefaultPerfCallback);
   }

5. Call the invoke() method in the loop() function to process the sensor data, then the registered callback function will be called (if the output category is registered in corresponding callback functions):

   void loop() {
       auto frame = capture.getManagedFrame();
       instance.invoke(frame);    
   }

Example

Please visit the examples folder to find sample sketches that demonstrate how to use the library with supported sensors.

API Reference

Constructors & Destructors

• SSCMAMicroCore()

  • Initializes a new instance of the SSCMAMicroCore class. This is where any necessary setup or initialization can occur.

• ~SSCMAMicroCore()

  • Cleans up any resources allocated by the SSCMAMicroCore instance. This is where any necessary cleanup or deinitialization should occur.

Configuration and Invocation

• begin(const Config& config)

  • This method initializes the SSCMAMicroCore with a specific configuration. It sets up the model and algorithm to be used for subsequent detections.
  • Returns: An Expected struct containing a boolean success indicating whether the operation was successful and a message providing additional information.

• invoke(const Frame& frame, const InvokeConfig* config = nullptr, void* user_context = nullptr)

  • This method processes a single frame using the SSCMAMicroCore. It can be called repeatedly to process multiple frames.
  • Parameters:
    • frame: The frame of image data to be processed.
    • config: An optional pointer to an InvokeConfig struct that overrides the configuration set in begin.
    • user_context: A user-provided pointer that will be passed to the registered callbacks.
  • Returns: An Expected struct containing a boolean success indicating whether the operation was successful and a message providing additional information.

• invoke(std::shared_ptr<Frame> frame, const InvokeConfig* config = nullptr, void* user_context = nullptr)

  • This method is similar to the previous invoke method, but it accepts a shared pointer to a managed frame instead of a reference.

Callback Registration

• void registerBoxesCallback(BoxesCallback callback)

  • Registers a callback function that will be called when box detection results are available.

• void registerClassesCallback(ClassesCallback callback)

  • Registers a callback function that will be called when class detection results are available.

• void registerPointsCallback(PointsCallback callback)

  • Registers a callback function that will be called when point detection results are available.

• void registerKeypointsCallback(KeypointsCallback callback)

  • Registers a callback function that will be called when keypoint detection results are available.

• void registerPerfCallback(PerfCallback callback)

  • Registers a callback function that will be called when performance metrics are available.

• const std::vector<Box>& getBoxes() const

  • Returns the box detection results.

• const std::vector<Class>& getClasses() const

  • Returns the classification results.

• const std::vector<Point>& getPoints() const

  • Returns the point detection results.

• const std::vector<Keypoints>& getKeypoints() const

  • Returns the keypoint detection results.

• const Perf& getPerf() const

  • Returns the performance metrics.

Please refer to the header files for detailed information about the library's types and APIs.

Compatibility

This library is compatible with Arduino boards and sensors that support the SSCMA-Micro firmware.

License

This library is released under the Apache License.

Contributions

Contributions and improvements to the library are welcomed. If you encounter any issues or have suggestions for enhancements, please create an issue or pull request here.

For more detailed information, including specific API references, please refer to the header files and the provided examples included in the library.

Thank you for using the Seeed_Arduino_SSCMACore library! We hope it enhances your experience in working with SSCMA-Micro Core with supported sensors on the Arduino platform.