Notebook 7.6 - Pedestrian Detection

In this notebook, you will use computer vision to detect pedestrians.

! pip install imutils opencv-python matplotlib

Requirement already satisfied: imutils in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (0.5.4)
Requirement already satisfied: opencv-python in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (4.5.4.60)
Requirement already satisfied: matplotlib in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (3.5.0)
Requirement already satisfied: numpy>=1.17.3 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from opencv-python) (1.21.4)
Requirement already satisfied: python-dateutil>=2.7 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (2.8.2)
Requirement already satisfied: pyparsing>=2.2.1 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (3.0.6)
Requirement already satisfied: kiwisolver>=1.0.1 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (1.3.2)
Requirement already satisfied: fonttools>=4.22.0 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (4.28.1)
Requirement already satisfied: packaging>=20.0 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (21.3)
Requirement already satisfied: pillow>=6.2.0 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (8.4.0)
Requirement already satisfied: cycler>=0.10 in c:\users\hc6118\anaconda3\envs\tsenv\lib\site-packages (from matplotlib) (0.11.0)
Requirement already satisfied: six>=1.5 in c:\users\hc6118\appdata\roaming\python\python38\site-packages (from python-dateutil>=2.7->matplotlib) (1.15.0)

import cv2
import imutils
import numpy as np
import matplotlib.pyplot as plt
from imutils.object_detection import non_max_suppression

Initialize descriptor and load pre-trained detector

hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())

Set path to the image and load it using OpenCV

image_path = 'data/example.jpg'

image = cv2.imread(image_path)

Plot the original figure

plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
plt.show()

Detect pedestrian in the image, performance highly affected by the parameters

(rects, weights) = hog.detectMultiScale(image, winStride=(6, 6), padding=(24, 24), scale=1.03)

Suppress overlapping bounding boxes

rects = np.array([[x, y, x + w, y + h] for (x, y, w, h) in rects])
merge_boxes = non_max_suppression(rects, probs=None, overlapThresh=0.65)

Draw bounding boxes on detected pedestrian

for (x1, y1, x2, y2) in merge_boxes:
    cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)

Plot the figure with bounding box showing the detected pedestrians

plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
plt.show()