import cv2
import numpy as np # type: ignore
import pandas as pd # type: ignore
import imutils
from backend.utils import timeit
DELTA_THRESH = 10
MIN_AREA = 4000
class Detector():
"""Class motion detector"""
@timeit
def __init__(self):
self.avg = None
self.colors = np.random.uniform(0, 255, size=(100, 3))
@timeit
def prediction(self, image):
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
image = cv2.GaussianBlur(image, (21, 21), 0)
if self.avg is None:
self.avg = image.copy().astype(float)
cv2.accumulateWeighted(image, self.avg, 0.5)
frameDelta = cv2.absdiff(image, cv2.convertScaleAbs(self.avg))
thresh = cv2.threshold(
frameDelta, DELTA_THRESH, 255,
cv2.THRESH_BINARY)[1]
thresh = cv2.dilate(thresh, None, iterations=2)
cnts = cv2.findContours(
thresh.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
cnts = imutils.grab_contours(cnts)
self.avg = image.copy().astype(float)
return cnts
@timeit
def filter_prediction(self, output, image):
if len(output) < 2:
return pd.DataFrame()
else:
df = pd.DataFrame(output)
df = df.assign(
area=lambda x: df[0].apply(lambda x: cv2.contourArea(x)),
bounding=lambda x: df[0].apply(lambda x: cv2.boundingRect(x))
)
df = df[df['area'] > MIN_AREA]
df_filtered = pd.DataFrame(
df['bounding'].values.tolist(), columns=['x1', 'y1', 'w', 'h'])
df_filtered = df_filtered.assign(
x1=lambda x: x['x1'].clip(0),
y1=lambda x: x['y1'].clip(0),
x2=lambda x: (x['x1'] + x['w']),
y2=lambda x: (x['y1'] + x['h']),
label=lambda x: x.index.astype(str),
class_name=lambda x: x.index.astype(str),
)
return df_filtered
def draw_boxes(self, image, df):
for idx, box in df.iterrows():
color = self.colors[int(box['label'])]
cv2.rectangle(image, (box['x1'], box['y1']), (box['x2'], box['y2']), color, 6)
cv2.putText(image, box['label'], (box['x1'], box['y1'] - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)
return image
if __name__ == "__main__":
image = cv2.imread("./imgs/image.jpeg")
print(image.shape)
detector = Detector()
image2 = cv2.imread("./imgs/image_box.jpg")
print(image2.shape)
image2 = cv2.cvtColor(image2, cv2.COLOR_BGR2GRAY)
image2 = cv2.GaussianBlur(image2, (21, 21), 0)
detector.avg = image2.astype(float)
output = detector.prediction(image)
df = detector.filter_prediction(output, image)
print(df)
image = detector.draw_boxes(image, df)
cv2.imwrite("./imgs/outputcv.jpg", image)
