> ## Documentation Index
> Fetch the complete documentation index at: https://docs.efference.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Tutorials

> Runnable C++ programs, one per part of the SDK

The SDK ships a set of small, self-contained programs under
[`tutorials/`](https://github.com/EfferenceAI/efference/tree/main/tutorials),
one topic per directory. Each has its own source, README, and build script, and
each builds standalone against the installed SDK.

Build the SDK first, then build all tutorials or an individual one:

```sh theme={null}
./build.sh                              # from the repo root: SDK + tools
./build.sh --tutorials                  # every tutorial
tutorials/serial_number/cpp/build.sh    # or one, from its own folder
```

Each binary is named after its source file and lands in that tutorial's
`build/`:

```sh theme={null}
tutorials/serial_number/cpp/build/serial_number
```

## Wired (USB)

| Topic                 | What it shows                                                       |
| --------------------- | ------------------------------------------------------------------- |
| `serial_number`       | The minimal program: open over USB and read `DeviceInformation`.    |
| `health_check`        | A `health_check()` sweep, with `--deep` for the stress tier.        |
| `data_stream`         | The `grab()` to `retrieve_image()` / `retrieve_imu()` loop.         |
| `grab`                | A bare open/grab/retrieve loop for throughput checks and debugging. |
| `wifi_status`         | Reading the M1's WiFi association from the cached snapshot.         |
| `record_and_download` | Record to the device, then pull the MCAP over USB.                  |
| `ota_sideload`        | Update firmware from a local `.eff` over USB, with no network.      |

## Wireless (Bluetooth LE control, WiFi data)

| Topic                   | What it shows                                                                   |
| ----------------------- | ------------------------------------------------------------------------------- |
| `discover_devices`      | `get_device_list(scan_ble=true)`: find USB and BLE devices without opening one. |
| `choosing_a_connection` | The same code path over USB or Bluetooth LE.                                    |
| `wifi_provisioning`     | `wifi_add()` and `wifi_select()` over either transport.                         |
| `udp_livestream`        | The fully wireless data path: BLE control plus WiFi/UDP video and IMU.          |
| `record_and_upload`     | BLE control plus a device-driven upload to an S3 or HTTP URL.                   |
| `ota_update`            | Firmware update over the device's own WiFi.                                     |

## OpenCV

Built only where both OpenCV and FFmpeg are present. Both use `ef::toCvMat()`
from `<ef/OpenCV.hpp>` for a zero-copy view of a grabbed frame as a `cv::Mat`.

| Topic              | What it shows                                                 |
| ------------------ | ------------------------------------------------------------- |
| `opencv_display`   | `grab()` to `retrieve_image(VIEW::BGR)` to `cv::imshow()`.    |
| `calibrate_camera` | Live checkerboard capture and a Double Sphere intrinsics fit. |

## Calibration

| Topic           | What it shows                                                            |
| --------------- | ------------------------------------------------------------------------ |
| `calibrate_imu` | Gyro bias and accelerometer ellipsoid estimation, written to the device. |

Both calibration tutorials need the prebuilt calibration library for your
architecture and are skipped where it is unavailable. They write to the device,
so read [Calibration](/device/calibration) before running them.

## Walkthroughs

The pages that follow work through the most common of these inline, with the
wired and wireless variants side by side.