Edge-AI with ESP32-S3 Workshop: Assignment 1

Assignment 1: Install ESP-IDF and ESP-WHO

#

In this assignment you will set up everything you need to build and flash ESP-WHO applications on the ESP32-S3-EYE. By the end, you will have ESP-IDF installed and managed with EIM, and the ESP-WHO repository cloned and ready to use.

Step 1: Install ESP-IDF

#

The recommended way to install and manage ESP-IDF is through the ESP-IDF Installation Manager (EIM), available in both a graphical and a command-line edition. You can download EIM from the official page.

Alternatevely, you can download directly from the terminal:

brew tap espressif/eim
brew install eim

brew tap espressif/eim
brew install --cask eim-gui

echo "deb [trusted=yes] https://dl.espressif.com/dl/eim/apt/ stable main" | sudo tee /etc/apt/sources.list.d/espressif.list
sudo apt update
sudo apt install eim-cli

echo "deb [trusted=yes] https://dl.espressif.com/dl/eim/apt/ stable main" | sudo tee /etc/apt/sources.list.d/espressif.list
sudo apt update
sudo apt install eim

winget install Espressif.EIM-CLI

Once EIM is installed, use it to install ESP-IDF. Run the following command to install the version required by ESP-WHO:

eim install v5.5.4

Alternatively, if you prefer a guided setup, use the interactive wizard:

eim wizard

For a full step-by-step walkthrough of the installation process, refer to the dedicated setup workshop:

ESP-IDF Preliminary Setup Workshop

After installation, check which versions are available on your machine:

eim list

You should see output similar to:

Installed ESP-IDF versions:
  v5.5.4   ~/.espressif/frameworks/esp-idf-v5.5.4

If you have multiple versions installed, set the active one with:

eim select v5.5.4

This prints the activation command for your OS. Run it to make idf.py available in your terminal:

source ~/.espressif/tools/activate_idf_v5.5.4.sh

source ~/.espressif/tools/activate_idf_v5.5.4.sh

. $HOME\.espressif\tools\activate_idf_v5.5.4.ps1

%USERPROFILE%\.espressif\tools\activate_idf_v5.5.4.bat

Once the environment is active, verify it in your terminal:

idf.py --version

You should see output similar to:

ESP-IDF v5.5.4

Step 2: Clone ESP-WHO

#

Choose a suitable location for the ESP-WHO repository — for example, your home directory or a dedicated ~/esp/ workspace folder. Do not clone it inside an existing project folder, as ESP-WHO is a shared framework used across multiple projects in this workshop.

# Example: clone into ~/esp/
mkdir -p ~/esp && cd ~/esp
git clone --recursive https://github.com/espressif/esp-who.git

After cloning, check out the specific commit validated for this workshop:

cd esp-who
git checkout 1681a1ce7dd356dfa541138d4b25d2ae1395472f
git submodule update --init --recursive

This commit is the version of ESP-WHO that has been validated for the ESP32-S3-EYE with ESP-IDF v5.5.4. It includes the correct BSP configuration, component versions, and example code that match the instructions throughout this workshop.

Step 3: Set the ESP-WHO environment variable

#

ESP-WHO uses a custom idf.py action to simplify project configuration. You need to point the IDF_EXTRA_ACTIONS_PATH variable to the tools/ folder inside your ESP-WHO clone.

export IDF_EXTRA_ACTIONS_PATH=/path/to/esp-who/tools
echo $IDF_EXTRA_ACTIONS_PATH

export IDF_EXTRA_ACTIONS_PATH=/path/to/esp-who/tools
echo $IDF_EXTRA_ACTIONS_PATH

$Env:IDF_EXTRA_ACTIONS_PATH = "C:\path\to\esp-who\tools"
echo $Env:IDF_EXTRA_ACTIONS_PATH

set IDF_EXTRA_ACTIONS_PATH=C:\path\to\esp-who\tools
echo %IDF_EXTRA_ACTIONS_PATH%

Step 4: Verify the setup

#

Let’s do a quick build check using one of the ESP-WHO examples to confirm everything is wired up correctly. Navigate to the face recognition example:

cd esp-who/examples/human_face_recognition

Configure the project for the ESP32-S3-EYE board:

idf.py -DSDKCONFIG_DEFAULTS=sdkconfig.bsp.esp32_s3_eye set-target esp32s3

The -DSDKCONFIG_DEFAULTS flag tells CMake which default configuration file to use when generating sdkconfig. ESP-WHO examples ship with one sdkconfig.bsp.* file per supported board, each containing the correct pin assignments, clock speeds, and peripheral settings for that specific hardware. By passing sdkconfig.bsp.esp32_s3_eye, you ensure the project is configured for the ESP32-S3-EYE without having to set each option manually through menuconfig. You will use this flag every time you configure a new ESP-WHO example.

Build the project:

idf.py build

If the build completes without errors, your environment is ready.

Next step

#

You now have a working ESP-IDF and ESP-WHO setup. Time to run your first vision AI application.

Assignment 2: ESP-WHO - Working with face detection

Return to the workshop main page