在计算机视觉领域,行人检测是一个重要的研究方向,它在视频监控、自动驾驶、人机交互等领域都有着广泛的应用。本文将介绍如何使用 OpenCV 库来实现视频中的行人检测。
环境准备
首先,我们需要安装 OpenCV 库。可以通过以下命令来安装:
pip install opencv-python
代码实现
以下是实现视频中行人检测的代码:
import cv2
import os
import numpy as np
import time
def detect_people():
current_dir = os.path.dirname(os.path.abspath(__file__))
video_path = os.path.join(current_dir, 'walk2.mp4')
if not os.path.exists(video_path):
print(f"Error: 视频文件不存在,请确认文件路径: {video_path}")
return
# 创建HOG检测器
hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())
cap = cv2.VideoCapture(video_path)
if not cap.isOpened():
print("Error: 无法打开视频文件")
return
# 设置视频捕获的分辨率
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 640)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)
# 用于控制检测频率
frame_count = 0
detection_interval = 3 # 每隔3帧进行一次检测
last_boxes = []
while True:
start_time = time.time()
ret, frame = cap.read()
if not ret:
break
# 降低分辨率
frame = cv2.resize(frame, (640, 480))
# 每隔几帧进行一次检测
if frame_count % detection_interval == 0:
# 检测人
boxes, weights = hog.detectMultiScale(frame,
winStride=(8, 8),
padding=(4, 4),
scale=1.1)
last_boxes = boxes
else:
boxes = last_boxes
# 在图像上绘制边界框
for (x, y, w, h) in boxes:
cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
# 显示人数
people_count = len(boxes)
cv2.putText(frame, f'People Count: {people_count}',
(10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1,
(0, 255, 0), 2)
# 计算和显示FPS
fps = 1.0 / (time.time() - start_time)
cv2.putText(frame, f'FPS: {int(fps)}',
(10, 60), cv2.FONT_HERSHEY_SIMPLEX, 1,
(0, 255, 0), 2)
# 显示结果
cv2.imshow('People Detection', frame)
frame_count += 1
# 减小等待时间,提高帧率
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
if __name__ == '__main__':
detect_people()