.. _advance_tutorial:

4. Advanced
========================

Environment Setup
-------------------

Before running any project in this chapter, you need to set up a Python virtual environment and install the required dependencies.

#. Open PowerShell or CMD, navigate to the downloaded code directory, then enter the ``Advanced`` folder.

#. Run the setup script:

   .. code-block:: bash

      .\setup.bat

   This will automatically create a ``venv`` virtual environment and install all required pip packages inside it (your system Python environment will not be affected).

#. After installation completes, activate the virtual environment:

   - PowerShell: ``.\venv\Scripts\Activate.ps1``
   - CMD: ``.\venv\Scripts\activate.bat``

   Once activated, you will see ``(venv)`` at the beginning of the command line.

#. Navigate to the corresponding project directory and run the script:

   .. code-block:: bash

      python <script_name>.py

Part 1: OpenCV
---------------------

.. note::
   All OpenCV projects require the UVC webcam firmware. Please flash the ``usb_webcam`` project from the :ref:`ESP-IDF chapter <esp_idf>` to the development board first.

.. note::
   Make sure the virtual environment is activated before running any project.

For each project below, open VS Code, navigate to the ``Advanced/1.openCV`` directory, right-click and select "Open in Integrated Terminal".

1. Contour Detection
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Captures images from the UVC camera and uses OpenCV to detect object contours.

Steps:

#. Connect the camera to your computer and verify it works using the built-in Camera app.
#. Run the demo:

   .. code-block:: bash
                                                                                                                                          
      python openCV.py

#. You will see the camera feed with green rectangular outlines drawn around detected object edges.

.. image:: img/advanced/opencv.jpg

2. Color Tracking
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Captures images from the UVC camera and uses OpenCV for color calibration and tracking.

Steps:

#. Connect the camera to your computer and verify it works using the built-in Camera app.
#. Run the demo:

   .. code-block:: bash

      python color_tracker.py

#. Two windows will appear. The first window has a crosshair in the center. Aim it at the object whose color you want to track and press ``Space``. OpenCV will then track objects of that color and overlay the result in the first window.

.. image:: img/advanced/opencv_color.png

3. Face Detection
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Captures images from the UVC camera and uses OpenCV for real-time face detection.

Steps:

#. Connect the camera to your computer and verify it works using the built-in Camera app.
#. Run the demo:

   .. code-block:: bash

      python face_detector.py --model models/res10_300x300_ssd_iter_140000.caffemodel --config models/deploy.prototxt --conf 0.6

#. A video window will appear showing detected faces.

.. image:: img/advanced/opencv_face.png

Part 2: YOLO
---------------------

.. note::
   All YOLO projects require the UVC webcam firmware. Please flash the ``usb_webcam`` project from the :ref:`ESP-IDF chapter <esp_idf>` to the development board first.

For each project below, open VS Code, navigate to the ``Advanced/2.yolo`` directory, right-click and select "Open in Integrated Terminal".

1. Object Recognition (YOLOv5su)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Uses YOLOv5su to perform real-time object detection and classification from the UVC camera feed.

Steps:

#. Connect the camera to your computer and verify it works using the built-in Camera app.
#. Run the demo:

   .. code-block:: bash

      python 1.yolo_object_recognition.py

#. A video stream window will appear. Point the camera at objects and YOLOv5su will identify and classify them in real time.

.. image:: img/advanced/yolo1.png

2. Pose Estimation (YOLOv8n-pose)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Uses YOLOv8n-pose to perform real-time human body pose estimation from the UVC camera feed.

Steps:

#. Connect the camera to your computer and verify it works using the built-in Camera app.
#. Run the demo:

   .. code-block:: bash

      python 2.yolo_pose.py

#. A video stream window will appear. Point the camera at a person and the model will detect and display body joint keypoints in real time.

.. image:: img/advanced/yolo2.png