def map_time_step(img_path,
rpm=19.6,
field=14,
diameter=70,
suffix=".bmp",
save_path=None):
img_list = sorted(gb.glob(img_path + r"/*" + suffix), key=os.path.getmtime)
pvoc = PascalVocXmlParser()
# Get the linear velocity of the rotating cloth
perimeter = np.pi * diameter
min_velo = rpm * perimeter
speed = min_velo / 60.0
for i in range(1, len(img_list), 1):
img_file = img_list[i]
print(img_file)
ann_file = get_ann_file(img_file)
if os.path.isfile(ann_file):
boxes = pvoc.get_boxes(ann_file)
else:
boxes = []
pre_img_file = img_list[i - 1]
pre_ann_file = get_ann_file(pre_img_file)
if os.path.isfile(pre_ann_file):
pre_boxes = pvoc.get_boxes(pre_ann_file)
else:
pre_boxes = []
img = cv2.imread(img_file, cv2.IMREAD_COLOR)
img_h, img_w = img.shape[:2]
intv = get_time_interval(img_file, pre_img_file)
nboxes = map_next_boxes(pre_boxes, speed, intv, field, (img_h, img_w))
#print(boxes)
#print(nboxes)
img = draw_boxes(img, boxes, color=(0, 0, 255))
img = draw_boxes(img, nboxes, color=(255, 0, 0))
if save_path is not None:
_, filename = os.path.split(img_file)
cv2.imwrite(os.path.join(save_path, filename), img)
else:
cv2.imshow("image", img)
cv2.waitKey(0)
import os, sys
import numpy as np
import glob as gb
import xml.etree.ElementTree as ET
from sklearn.cluster import KMeans
from matplotlib import pyplot as plt
from PascalVocParser import PascalVocXmlParser
pvoc = PascalVocXmlParser()
def show_histogram(path):
dataset = encode_dataset(path)
widths = dataset[:, 0]
heights = dataset[:, 1]
plt.subplot(1, 2, 1), plt.hist(widths), plt.title("Width Distribution")
plt.subplot(1, 2, 2), plt.hist(heights), plt.title("Heights Distribution")
plt.show()
def cluster_anchors(path, k=6, label=None):
dataset = encode_dataset(path, label=label)
km = KMeans(n_clusters=k, random_state=9).fit(dataset)
ac = np.array(km.cluster_centers_, dtype=np.int32)
return ac
def encode_dataset(path, label=None):
dataset = [] # dataset that x_length larger than x_thres
for xml_file in gb.glob(path + r"/*.xml"):
def __init__(self, params):
self.pvoc = PascalVocXmlParser()
self.updateParams(params)
class CheckLabels(object):
def __init__(self, params):
self.pvoc = PascalVocXmlParser()
self.updateParams(params)
def updateParams(self, params):
self.format = params['format']
self.is_alert = params['is_alert']
self.collection = params['collection']
self.params = params
self.matrix = {}
self.abnormal_matrix = {}
def __call__(self, input):
if os.path.isfile(input):
_, suffix = os.path.splitext(input)
if suffix[1:] == 'json':
self.checkJson(input)
elif suffix[1:] == 'xml':
self.checkXml(input)
else:
raise ValueError('Unsupported input format.')
elif os.path.exists(input):
self.checkDir(input)
else:
raise ValueError('Invalid input.')
def checkDir(self, dir):
if self.format.lower() in ['labelme', 'json']:
file_list = gb.glob(dir + r'/*.json')
for json_file in file_list:
self.checkJson(json_file, disp=False)
elif self.format.lower() in ['labelimg', 'xml', 'pascalvoc', 'voc']:
file_list = gb.glob(dir + r'/*.xml')
for xml_file in file_list:
self.checkXml(xml_file, disp=False)
else:
raise ValueError('Invalid format.')
self._display()
def checkJson(self, json_file, disp=True):
print('Processing file', json_file, '...')
with open(json_file, "r", encoding="utf-8") as f:
js_obj = json.load(f)
for elem in js_obj['shapes']:
self._register(elem['label'])
if self.is_alert:
self._register_abnormal(elem['label'], json_file)
f.close()
if disp: self._display()
def checkXml(self, xml_file, disp=True):
print('Processing file', xml_file, '...')
labels = self.pvoc.get_labels(xml_file)
for label in labels:
self._register(label)
if self.is_alert:
self._register_abnormal(label, xml_file)
if disp: self._display()
def _register(self, label):
if label not in self.matrix:
self.matrix[label] = 1
else:
self.matrix[label] += 1
def _register_abnormal(self, label, ann_file):
if label in self.collection: return
_, filename = os.path.split(ann_file)
if label not in self.abnormal_matrix:
self.abnormal_matrix[label] = [filename]
else:
self.abnormal_matrix[label].append(filename)
def _display(self):
print('\nDisplaying the results:\n')
for label in self.matrix:
print(label + ': ' + str(self.matrix[label]))
if not self.is_alert: return
for ab_label in self.abnormal_matrix:
print('Abnormal label: ' + ab_label + '\n')
for filename in self.abnormal_matrix[ab_label]:
print(' ' + filename + '\n')