Me and my project members are trying do develop a small application for traffic sign recognition with OpenCV and python. Our goal is that each traffic sign can be detected by realtime (webcam) or video file with a low percentage of errors for our student project. We decided to work with the haarlike feature to train different signs such as "stop-sign", "priority-sign" and so on. The result is a XML-File for each specific sign which can be used to detect.
The problem is now to combine these XML-files in one python script. We also saw the "facedetection.py" which shows only two haar classifiers which are interleaved and only spotted in one big green rectangle and inside this is another blue one for eye detection.
The question is if it´s possible to combine the 30 traffic sign XML-files in a python script, by using threads or other solutions?
I´m a newbie in python so I tried to combine three classifier which are independently to each other, but the result is really bad because the more xml-files you implement the less is the speed/benchmark of the program.
#!/usr/bin/env python
# Python 2/3 compatibility
from __future__ import print_function
import numpy as np
import cv2
# local modules
from video import create_capture
from common import clock, draw_str
def detect(img, cascade):
rects = cascade.detectMultiScale(img, scaleFactor=1.3, minNeighbors=4, minSize=(30, 30),
flags=cv2.CASCADE_SCALE_IMAGE)
if len(rects) == 0:
return []
rects[:,2:] += rects[:,:2]
return rects
def draw_rects(img, rects, color):
for x1, y1, x2, y2 in rects:
cv2.rectangle(img, (x1, y1), (x2, y2), color, 2)
if __name__ == '__main__':
import sys, getopt
print(__doc__)
args, video_src = getopt.getopt(sys.argv[1:], '', ['cascade=', 'nested-cascade=', 'my-cascade='])
try:
video_src = video_src[0]
except:
video_src = 0
args = dict(args)
cascade_fn = args.get('--cascade', "/Users/daniel/opencv/data/haarcascades/haarcascade_frontalface_default.xml")
nested_fn = args.get('--nested-cascade', "/Users/daniel/opencv/data/haarcascades/haarcascade_eye.xml")
my_cascade = args.get('--my-cascade', "/Users/daniel/Downloads/cascade_speed20_server.xml")
cascade = cv2.CascadeClassifier(cascade_fn)
nested = cv2.CascadeClassifier(nested_fn)
mycascade = cv2.CascadeClassifier(my_cascade)
cam = create_capture(video_src, fallback='synth:bg=../data/lena.jpg:noise=0.05')
while True:
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects1 = detect(gray, cascade)
vis1 = img.copy()
draw_rects(vis1, rects1, (255, 255, 0))
rects2 = detect(gray, nested)
vis2 = img.copy()
draw_rects(vis2, rects2, (0, 255, 0))
rects3 = detect(gray, mycascade)
vis3 = img.copy()
draw_rects(vis3, rects3, (0, 255, 255))
dt = clock() - t
draw_str(vis1, (20, 20), 'time: %.1f ms' % (dt*1000))
draw_str(vis2, (20, 20), 'time: %.1f ms' % (dt*1000))
draw_str(vis3, (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('facedetect', vis1)
cv2.imshow('facedetect', vis2)
cv2.imshow('facedetect', vis3)
if 0xFF & cv2.waitKey(5) == 27:
break
cv2.destroyAllWindows()
