ESP32-CAM is an inexpensive MJPEG stream embedded device. The library is MJPEG stream decoder based on libcurl and OpenCV, and written in C/C++.

Modify the following code in "CameraWebServer.ino" file, that is MJPEG stream boundary.

#define PART_BOUNDARY "WINBONDBOUDARY"
static const char* _STREAM_CONTENT_TYPE = "multipart/x-mixed-replace;boundary=" PART_BOUNDARY;
static const char* _STREAM_BOUNDARY = "\r\n--" PART_BOUNDARY "\r\n";
static const char* _STREAM_PART = "Content-Type: image/jpeg\r\n\r\n";

• OpenCV 3 or 4

• libcurl 7

• wxWidgets 2.8.12(for wxESP32-CAM example)

Include "ESP32-CAM MJPEG Library" folder into your project.

• ESP32-CAM Library.h
• ESP32-CAM Library.cpp

#include "./ESP32-CAM MJPEG Library/ESP32-CAM Library.h"
typedef enum {
 FRAMESIZE_QQVGA, // 160x120
 FRAMESIZE_QQVGA2, // 128x160
 FRAMESIZE_QCIF, // 176x144
 FRAMESIZE_HQVGA, // 240x176
 FRAMESIZE_QVGA, // 320x240
 FRAMESIZE_CIF, // 400x296
 FRAMESIZE_VGA, // 640x480
 FRAMESIZE_SVGA, // 800x600
 FRAMESIZE_XGA, // 1024x768
 FRAMESIZE_SXGA, // 1280x1024
 FRAMESIZE_UXGA // 1600x1200
} framesize_t;
ESP32_CAM *esp32_cam = new ESP32_CAM(std::string( "192.168.1.254" )); //ESP32-CAM local IP address
esp32_cam->SetResolution(FRAMESIZE_SVGA); // Set mjpeg video stream resolution
esp32_cam->StartVideoStream(); // Start mjpeg video stream thread
//cv::Mat frame = esp32_cam->GetFrame(); // Get mjpeg frame
cv::Mat frame; 
esp32_cam >> frame; // "operator >>" call GetFrame(), like cv::VideoCapture call read() 
if(!frame.empty()){
 cv::imshow("Example",frame); //Show mjpeg video stream
}

In cmd_handler, "var" is variable name, "val" is value of variable, curl POST command is:

int ESP32_CAM::SendStream(CURL *curl_object)
{
 long return_code;
 CURLcode response;
 curl_easy_setopt(curl_object,CURLOPT_URL,"192.168.1.254/control?var=framesize&val=7");
 response = curl_easy_perform(curl_object);
 response = curl_easy_getinfo(curl_object, CURLINFO_RESPONSE_CODE, &return_code);
 return (int)return_code;
}

"192.168.1.254/control?var=framesize&val=8", "192.168.1.254" is ESP32-CAM IP address setting from:

// Set your Static IP address
IPAddress local_IP(192, 168, 1, 254);
// Set your Gateway IP address
IPAddress gateway(192, 168, 1, 1);
WiFi.begin(ssid, password); 
WiFi.config(local_IP, gateway, subnet);

In this example, "/control?var=framesize&val=8" POST /control API and set parameter "framesize" value is "7"(FRAMESIZE_SVGA 800x600).

API FlashControl(bool), Flash LED turn on or off "/led?var=flash&val=1", val=1 is on, val=0 is off.

Set curl_easy_setopt parameter CURLOPT_WRITEFUNCTION and CURLOPT_WRITEDATA for writing received data.

std::string ESP32_CAM::GetRSSI()
{
 std::string buffer;
 curl_easy_setopt(curl_command,CURLOPT_WRITEFUNCTION,WriteCallback);
 curl_easy_setopt(curl_command,CURLOPT_WRITEDATA,&buffer); 
 if(SendStream(curl_command,std::string("/RSSI"))){
 } 
 return buffer;
}

alt text

alt text

In this case MJPEG stream boundary is fixed length, for split into a MJPEG stream of JPEG binary(byte), each frame JPEG binary with cv::imdecode decode to RGB format.

const char VIDEO_STREAM_INTERLEAVE[] = "--WINBONDBOUDARY\r\nContent-Type: image/jpeg\r\n\r\n";
const char JPEG_SOI_MARKER_FIRST_BYTE = 0xFF;
const char JPEG_SOI_MARKER_SECOND_BYTE = 0xD8;

Content-type: multipart/x-mixed-replace; boundary=WINBONDBOUDARY
--WINBONDBOUDARY\r\n
Content-Type: image/jpeg\r\n\r\n
JPEG Binary(0xFF,0xD8 ... 0xFF,0xD9)
--WINBONDBOUDARY\r\n 
Content-Type: image/jpeg\r\n\r\n 
JPEG Binary(0xFF,0xD8 ... 0xFF,0xD9) 
...
...
...
--WINBONDBOUDARY\r\n 
Content-Type: image/jpeg\r\n\r\n 
JPEG Binary(0xFF,0xD8 ... 0xFF,0xD9) 

• "cd /examples/MAC/wxESP32-CAM/"
• make

alt text

Build x64 version wxWidgets and libcurl

• 『x64 Native Tools Command Prompt for VS 2017/2019』, open the program as an administrator.
• Build x64 libcurl 7.67.0
  • cd "./libcurl-7.67.0/winbuild"
  • run "buildconf.bat"
  • nmake /f Makefile.vc mode=static VC=15 MACHINE=x64 DEBUG=no
  • static link libcurl_a.lib, Ws2_32.lib, Wldap32.lib, winmm.lib, Crypt32.lib, Normaliz.lib
• Build x64 wxWidgets 3.1.2
  • cd "build/msw"
  • nmake /f makefile.vc /ls RUNTIME_LIBS=static SHARED=0 COMPILER_VERSION=141 TARGET_CPU=X64 BUILD=release
  • nmake /f makefile.vc /ls RUNTIME_LIBS=static SHARED=1 COMPILER_VERSION=141 TARGET_CPU=X64 BUILD=release
• Build wxESP32-CAM
  • Open *.sln project
  • Select "Release x64"
  • Rebuild

alt text

This example is wxWidgets GUI for the library demo, and integrate DNN Computer Vision use cases(YOLO V3, OpenPose).

2025年09月06日

• Google Drive
  • https://drive.google.com/drive/folders/1Ni5po9_sCTo7nfBxEJvBWAE7aVPQ1dz4
    
  • https://huggingface.co/camenduru/openpose/tree/f5bb0c0a16060ac8b373472a5456c76bd68eb202
    
  • https://huggingface.co/spaces/Epitech/Scarecrow/tree/main
    

Broken Links

• YOLO V3

  • Weights: https://pjreddie.com/media/files/yolov3.weights
    
  • Model: https://github.com/pjreddie/darknet/blob/master/cfg/yolov3.cfg?raw=true
    
  • COCO: https://github.com/pjreddie/darknet/blob/master/data/coco.names?raw=true
    

• OpenPose Caffe Model

  • https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/models/pose/mpi/pose_deploy_linevec.prototxt
    
  • http://posefs1.perception.cs.cmu.edu/Users/ZheCao/pose_iter_146000.caffemodel
    

OpenCV highgui MJPEG stream player.

• cd /examples/MAC/ESP32-CAM Opencv Example/
• make

alt text

• Open *.sln project
• Select "Release x64"
• Rebuild

alt text

• https://github.com/GCY/wxRovio

• https://github.com/tobybreckon/roviolib

• https://pjreddie.com/darknet/yolo/

• https://github.com/spmallick/learnopencv/tree/master/ObjectDetection-YOLO

• https://github.com/CMU-Perceptual-Computing-Lab/openpose

• https://github.com/spmallick/learnopencv/tree/master/OpenPose-Multi-Person

• https://randomnerdtutorials.com/esp32-cam-video-streaming-face-recognition-arduino-ide/

The following image shows the pinout diagram for the ESP32-CAM AI-Thinker. alt text

The following figure shows the schematic diagram for the ESP32-CAM.

alt text

Copyright (C) 2019 TonyGUO https://github.com/GCY.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.