ESP32-S3 AI CAM——基于yoloV5的AI安防系统
本帖最后由 kylinpoet 于 2025-5-6 14:04 编辑一、项目缘起：
ESP32 S3 + CAM 实在是低成本、高效的安防搭档。本项目通过在esp32上搭建http视频流服务，利用上位机实时获取摄像头的传递数据，进行识别。相应的数据处理在上位机完成。包括使用yolov5的预训练80分类数据进行人体识别，以及通过消息平台进行报警。
二、操作步骤：
1. 视频图传
在 dfrobot 的wiki库里找到并刷入，视频图传演示代码（点击访问），需要注意的是使用 Arduino 1.8版本刷入才行，Arduino 2 会输入失败（留待大佬解决）
#include "esp_camera.h"
#include <WiFi.h>
//
// WARNING!!! PSRAM IC required for UXGA resolution and high JPEG quality
// Ensure ESP32 Wrover Module or other board with PSRAM is selected
// Partial images will be transmitted if image exceeds buffer size
//
// You must select partition scheme from the board menu that has at least 3MB APP space.
// Face Recognition is DISABLED for ESP32 and ESP32-S2, because it takes up from 15
// seconds to process single frame. Face Detection is ENABLED if PSRAM is enabled as well
#define PWDN_GPIO_NUM -1
#define RESET_GPIO_NUM -1
#define XCLK_GPIO_NUM 5
#define Y9_GPIO_NUM 4
#define Y8_GPIO_NUM 6
#define Y7_GPIO_NUM 7
#define Y6_GPIO_NUM 14
#define Y5_GPIO_NUM 17
#define Y4_GPIO_NUM 21
#define Y3_GPIO_NUM 18
#define Y2_GPIO_NUM 16
#define VSYNC_GPIO_NUM 1
#define HREF_GPIO_NUM 2
#define PCLK_GPIO_NUM 15
#define SIOD_GPIO_NUM8
#define SIOC_GPIO_NUM9
// ===========================
// Enter your WiFi credentials
// ===========================
const char *ssid = "**********";
const char *password = "**********";
void startCameraServer();
void setupLedFlash(int pin);
void setup() {
Serial.begin(115200);
Serial.setDebugOutput(true);
Serial.println();
camera_config_t config;
config.ledc_channel = LEDC_CHANNEL_0;
config.ledc_timer = LEDC_TIMER_0;
config.pin_d0 = Y2_GPIO_NUM;
config.pin_d1 = Y3_GPIO_NUM;
config.pin_d2 = Y4_GPIO_NUM;
config.pin_d3 = Y5_GPIO_NUM;
config.pin_d4 = Y6_GPIO_NUM;
config.pin_d5 = Y7_GPIO_NUM;
config.pin_d6 = Y8_GPIO_NUM;
config.pin_d7 = Y9_GPIO_NUM;
config.pin_xclk = XCLK_GPIO_NUM;
config.pin_pclk = PCLK_GPIO_NUM;
config.pin_vsync = VSYNC_GPIO_NUM;
config.pin_href = HREF_GPIO_NUM;
config.pin_sccb_sda = SIOD_GPIO_NUM;
config.pin_sccb_scl = SIOC_GPIO_NUM;
config.pin_pwdn = PWDN_GPIO_NUM;
config.pin_reset = RESET_GPIO_NUM;
config.xclk_freq_hz = 20000000;
config.frame_size = FRAMESIZE_UXGA;
config.pixel_format = PIXFORMAT_JPEG;// for streaming
//config.pixel_format = PIXFORMAT_RGB565; // for face detection/recognition
config.grab_mode = CAMERA_GRAB_WHEN_EMPTY;
config.fb_location = CAMERA_FB_IN_PSRAM;
config.jpeg_quality = 12;
config.fb_count = 1;
// if PSRAM IC present, init with UXGA resolution and higher JPEG quality
// for larger pre-allocated frame buffer.
if (config.pixel_format == PIXFORMAT_JPEG) {
if (psramFound()) {
config.jpeg_quality = 10;
config.fb_count = 2;
config.grab_mode = CAMERA_GRAB_LATEST;
} else {
// Limit the frame size when PSRAM is not available
config.frame_size = FRAMESIZE_SVGA;
config.fb_location = CAMERA_FB_IN_DRAM;
}
} else {
// Best option for face detection/recognition
config.frame_size = FRAMESIZE_240X240;
#if CONFIG_IDF_TARGET_ESP32S3
config.fb_count = 2;
#endif
}
#if defined(CAMERA_MODEL_ESP_EYE)
pinMode(13, INPUT_PULLUP);
pinMode(14, INPUT_PULLUP);
#endif
// camera init
esp_err_t err = esp_camera_init(&config);
if (err != ESP_OK) {
Serial.printf("Camera init failed with error 0x%x", err);
return;
}
sensor_t *s = esp_camera_sensor_get();
// initial sensors are flipped vertically and colors are a bit saturated
if (s->id.PID == OV3660_PID) {
s->set_vflip(s, 1); // flip it back
s->set_brightness(s, 1); // up the brightness just a bit
s->set_saturation(s, -2);// lower the saturation
}
// drop down frame size for higher initial frame rate
if (config.pixel_format == PIXFORMAT_JPEG) {
s->set_framesize(s, FRAMESIZE_QVGA);
}
#if defined(CAMERA_MODEL_M5STACK_WIDE) || defined(CAMERA_MODEL_M5STACK_ESP32CAM)
s->set_vflip(s, 1);
s->set_hmirror(s, 1);
#endif
#if defined(CAMERA_MODEL_ESP32S3_EYE)
s->set_vflip(s, 1);
#endif
// Setup LED FLash if LED pin is defined in camera_pins.h
#if defined(LED_GPIO_NUM)
setupLedFlash(LED_GPIO_NUM);
#endif
WiFi.begin(ssid, password);
WiFi.setSleep(false);
Serial.print("WiFi connecting");
while (WiFi.status() != WL_CONNECTED) {
delay(500);
Serial.print(".");
}
Serial.println("");
Serial.println("WiFi connected");
startCameraServer();
Serial.print("Camera Ready! Use 'http://");
Serial.print(WiFi.localIP());
Serial.println("' to connect");
}
void loop() {
// Do nothing. Everything is done in another task by the web server
delay(10000);
}刷入成功后，我们会在串口中看到wifi连接信息。访问相应地址后，我们会看到以下内容。
这里要注意下，网址首页打开是视频流的形式，访问 /capture 子页面，是抓取的实时图片。
我们在上位机上用的就是这个子页面。
static/image/hrline/3.gif
2. 基于电脑yoloV5实现
yoloV5的代码，我们主要参考这里的文档（点击访问），但要做一些改变：
这里主要是判断是否检测到人类，以及为了正确阈值，设置的置信度。
另外还要增加发送消息预警的函数，这里是使用免费微信消息推送服务：server酱，当然你可以使用其它类似服务：
以下是识别到人类活动后的消息推送：
参考代码如下：
import cv2
import torch
import time
from yolov5 import YOLOv5
import urllib.request
import numpy as np
import requests
from img_upload import img_upload
url = 'http://192.168.1.43:80/capture'# ESP32-CAM的IP地址
sendkey = ''
# Load the YOLOv5 model
model_path = "d:/yolov5s.pt"# 模型路径
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model = YOLOv5(model_path, device=device)
last_alert_time = 0
# 通过server酱发送微信消息推送
def send_alert(title, img):
url = f"https://sctapi.ftqq.com/{sendkey}.send"
# 使用Markdown语法在desp中嵌入图片
image_url = img_upload(img)
desp = f""# 图片需为可访问的URL
payload = {
"text": title,
"desp": desp
}
response = requests.post(url, data=payload)
return response.json()
while True:
img_resp = urllib.request.urlopen(url)# 从URL获取图像数据
img_source = img_resp.read()
imgnp = np.array(bytearray(img_source), dtype=np.uint8)# 将图像数据转换为NumPy数
frame = cv2.imdecode(imgnp, -1)# 解码JPEG图像数据
# 记录处理开始时间
start_time = time.time()
# 将帧传递给模型进行预测
results = model.predict(frame, size=640)# 调整输入图像大小为640x640
# 遍历检测结果
for i, det in enumerate(results.pred):
if det is not None and len(det):
det = det.unsqueeze(0)
for *xyxy, conf, cls in det:
# 获取类别名称和置信度
class_name = results.names
confidence = float(conf)
print(f"检测到类别: {class_name}, 置信度: {confidence:.2f}")
if class_name == "person" and confidence > 0.5:
current_time = time.time()
# 检查是否距离上次报警超过5分钟（300秒）
if current_time - last_alert_time >= 300: # 5分钟内只报一次警
print(f"检测到人类活动，触发报警")
send_alert('检测到人类活动，触发报警', img_source)
last_alert_time = current_time# 更新上次报警时间
# 如需保存结果，可写入文件
# 获取预测结果并在图像上绘制
annotated_frame = results.render()# YOLOv5库中的render方法
# 记录处理结束时间
end_time = time.time()
# 计算帧率
processing_time = end_time - start_time
fps = 1 / processing_time
# 在图像上绘制帧率
# cv2.putText(annotated_frame, f"FPS: {fps:.2f}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2, cv2.LINE_AA)
# 提取可写副本
img_output = annotated_frame.copy()
cv2.putText(img_output, f"FPS: {fps:.2f}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2, cv2.LINE_AA)
# 显示带有预测结果的帧
cv2.imshow("YOLOv5 Detection", annotated_frame)
# 打印详细信息到终端
print(f"Processed frame in {processing_time:.4f} seconds, FPS: {fps:.2f}")
# 按下 'q' 键退出
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# 释放摄像头资源并关闭所有窗口
# cap.release()
cv2.destroyAllWindows()
顺便提下，yoloV5的预训练分类如下展示，class 0 是 person，其它可查看以下部分分类信息。完整内容请自行查阅。
3. 实践过程中的几点思考：
这里重点说明下测试过程中，填的几个坑：
1). 如上图①所示，示例代码里的模型路径，需要使用绝对路径，否则默认寻找的路径是 %homepath% ，也就是说需要将 pt 文件拷贝到这个目录，也就用户目录如：C:\Users\你的用户名\yolov5s.pt。如果此路径没有这个文件，他会去 https://github.com/ultralytics/y ... oad/v7.0/yolov5s.pt下载，而这个下载地址有时候是需要特殊网络的！！！
2). 如上图②所示，因为我装的是 GPU 版的torch！！！，这里应该是使用 'cuda:0'，而不是实例代码的 'cuda'。
三、安装图
页:
[1]