Skip to main content

Human Face Detection

Overview​

This example implements human face detection based on deep learning, providing developers with an extensible foundational framework for developing various face recognition-related applications.

The project is developed using the ESP-DL deep learning library, which offers powerful deep learning interfaces for ESP-WHO, enabling rich AIoT application scenarios when combined with various peripherals.

Preparation​

Hardware Requirements​

  • Development board supporting camera modules (recommended ESP32-S3 series)

Software Preparation​

1. Precompiled Firmware​

You can directly download the latest precompiled firmware for a quick experience:

2. Source Code Development Environment​

For further development, prepare the following environment:

  • Development Tool: Visual Studio Code (version 1.99.2 or higher)
  • Development Framework: ESP-IDF plugin (version 5.4.0 or higher)
  • Example Code: camthink-ai/esp-who

Important Note: Ensure to complete the environment setup according to the Development Environment Configuration Guide.

Function Verification​

Precompiled Firmware Testing Process​

1. Device Connection​

Use a Type-C data cable to connect the development board to your PC.

NE101 Main Board Interface Diagram

2. Firmware Flashing​

Refer to the detailed flashing guide:

3. Serial Output Parsing​

The device will output face detection results via serial port (baud rate: 115200):

Example output:

WhoDetect: 0, bbox: [0.826712, 47, 40, 169, 195], left_eye: [80, 105], left_mouth: [90, 151], nose: [101, 125], right_eye: [123, 99], right_mouth: [126, 146]

Data Field Explanation:

  • bbox: Face bounding box information
    • Format: [confidence, top-left x-coordinate, top-left y-coordinate, width, height]
  • left_eye: Left eye coordinates (x, y)
  • right_eye: Right eye coordinates (x, y)
  • nose: Nose coordinates (x, y)
  • left_mouth: Left mouth corner coordinates (x, y)
  • right_mouth: Right mouth corner coordinates (x, y)

4. Function Verification​

  1. Aim the camera at a face or a face photo.
  2. Observe the face feature point data output by the serial port.
  3. Try moving the camera to verify dynamic detection effects.

Static Face Detection Diagram

Dynamic Face Detection Effect Diagram

Source Code Development Verification Process​

1. Obtain the Source Code​

git clone https://github.com/camthink-ai/esp-who

2. Project Configuration​

  1. Open the project in VS Code: examples/human_face_detect

    Project Directory Structure

  2. Set the target chip to ESP32-S3.

    Chip Selection Interface

3. Compilation and Flashing​

  1. Compile the project.

    Compilation Interface

  2. Flash the firmware.

    Flashing Interface

4. Function Verification​

The verification method is the same as for the precompiled firmware, refer to the previous steps.

Reference Resources​

  1. ESP-DL Model Quantization Documentation:

  2. Official Example Repository: