|
|
|
""" |
|
Created on Tue Jan 19 13:24:25 2021 |
|
|
|
Código utilizado para gerar as máscaras e as imagens a partir dos videos e marcações XML geradas pelo sensarea. |
|
|
|
@author: Gabriel |
|
""" |
|
|
|
import cv2 |
|
import glob |
|
import os |
|
import pandas as pd |
|
import datetime |
|
import shapely |
|
import xml.etree.ElementTree as et |
|
from shapely.geometry import Polygon |
|
|
|
from shapely.ops import unary_union |
|
import numpy as np |
|
import cv2 |
|
|
|
|
|
|
|
|
|
def get_polygons(path): |
|
xtree = et.parse(path) |
|
xroot = xtree.getroot() |
|
|
|
masks = xroot.find("masks") |
|
df = pd.DataFrame(columns=['frame', 'polygon']) |
|
|
|
for mask in masks: |
|
|
|
dic = {} |
|
dic['frame'] = int(mask.find('frame').text) |
|
pontos = mask.find('polygon').attrib.get('points') |
|
|
|
tokens = pontos.split(' ') |
|
poly = [] |
|
for token in tokens[:-1]: |
|
ponto = token.split(',') |
|
poly.append((int(ponto[0]), int(ponto[1]))) |
|
|
|
dic['polygon'] = poly |
|
df = df.append(dic, ignore_index=True) |
|
|
|
return df |
|
|
|
|
|
|
|
def variance_of_laplacian(image): |
|
|
|
|
|
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) |
|
return cv2.Laplacian(image, cv2.CV_64F).var() |
|
|
|
path = "C:\\Users\\Gabriel\\OneDrive - Universidade de Tras-os-Montes e Alto Douro\\UTAD\\2020-2021\\Pesquisa\\Dataset\\Vídeos\\organizado\\dividido" |
|
path_xml = "C:\\Users\\Gabriel\\OneDrive - Universidade de Tras-os-Montes e Alto Douro\\UTAD\\2020-2021\\Pesquisa\\Dataset\\Vídeos\\mascaras" |
|
os.chdir(path) |
|
|
|
|
|
|
|
|
|
|
|
|
|
configs = [[512, int(512/2), 0.80, 0.5], [800, int(800/3), 0.80, 0.5]] |
|
|
|
|
|
next_frame = 30 |
|
|
|
|
|
dim_salvar = 512 |
|
tam_salvar = (dim_salvar, dim_salvar) |
|
|
|
columns = ['path', 'class', 'name', 'data_str', 'dia', 'mes', 'ano', 'set'] |
|
df=pd.DataFrame(columns=columns) |
|
parser = {'junho':'06', |
|
'mai':'05', |
|
'agosto':'08', |
|
'julho':'07', |
|
'set':'09', |
|
'maio':'05', |
|
'ago':'08'} |
|
|
|
for root, dirs, files in os.walk(path): |
|
for file in files: |
|
if file.endswith(".MTS"): |
|
dic = {} |
|
dic['path'] = os.path.join(root, file) |
|
dic['name'] = dic['path'].split(os.path.sep)[-1] |
|
dic['class'] = dic['path'].split(os.path.sep)[-2] |
|
dic['set'] = dic['path'].split(os.path.sep)[-3] |
|
df=df.append(dic, ignore_index=True) |
|
|
|
print(df.head()) |
|
|
|
|
|
path_to_save = 'C:\\Users\\Gabriel\\Downloads\\teste_masks_crop3' |
|
if not os.path.exists(path_to_save): |
|
os.mkdir(path_to_save) |
|
|
|
|
|
for i, row in df.iterrows(): |
|
|
|
for folder in ['image', 'mask']: |
|
|
|
if not os.path.exists(os.path.join(path_to_save, row['set'], folder)): |
|
|
|
os.makedirs(os.path.join(path_to_save, row['set'], folder), exist_ok=True) |
|
|
|
|
|
|
|
|
|
|
|
|
|
cap = cv2.VideoCapture(row['path']) |
|
|
|
|
|
length = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) |
|
|
|
|
|
path_polygon = os.path.join(path_xml, row['class'], row['name'].split('.')[0]+'.xml') |
|
|
|
|
|
df_polygon = None |
|
if os.path.isfile(path_polygon): |
|
df_polygon = get_polygons(os.path.join(path_xml, row['class'], row['name'].split('.')[0]+'.xml')) |
|
n_frame = 0 |
|
continua = True |
|
print('Video: ', row['path']) |
|
while(continua): |
|
|
|
res, frame = cap.read() |
|
|
|
if res: |
|
try: |
|
|
|
altura_imagem, largura_imagem = frame.shape[:2] |
|
|
|
|
|
|
|
frame_polygon = None |
|
|
|
if not df_polygon is None: |
|
frame_polygon = Polygon(df_polygon.iloc[n_frame]['polygon']).buffer(0) |
|
|
|
|
|
if frame_polygon.geom_type == 'MultiPolygon': |
|
|
|
frame_polygon =unary_union(list(frame_polygon)) |
|
if frame_polygon.geom_type == 'MultiPolygon': |
|
|
|
polys = list(frame_polygon) |
|
area = 0 |
|
index = None |
|
|
|
for i, pl in enumerate(polys): |
|
if pl.area > area: |
|
area = pl.area |
|
index = i |
|
|
|
frame_polygon = polys[index] |
|
|
|
|
|
|
|
|
|
mask = np.full((altura_imagem, largura_imagem), 0, dtype=np.uint8) |
|
|
|
|
|
|
|
coords = np.array([[x[0], x[1]] for x in list(frame_polygon.exterior.coords)], dtype=np.int32) |
|
|
|
coords = coords.reshape((-1,1,2)) |
|
|
|
mask = cv2.fillPoly(mask,[coords], 1) |
|
|
|
|
|
for c, config in enumerate(configs): |
|
|
|
image_size_side = config[0] |
|
|
|
stride = config[1] |
|
|
|
|
|
percentual_maximo = config[2] |
|
percentual_minimo = config[3] |
|
altura = 0 |
|
|
|
k = 0 |
|
|
|
|
|
while(altura+image_size_side < altura_imagem): |
|
|
|
largura = 0 |
|
|
|
while(largura+image_size_side < largura_imagem): |
|
|
|
crop_img = frame[altura:altura+image_size_side, largura:largura+image_size_side] |
|
|
|
|
|
crop_polygon = Polygon([(altura, largura), (altura, largura+image_size_side), (altura+image_size_side, largura+image_size_side), (altura+image_size_side, largura)]) |
|
|
|
|
|
crop_mask = mask[altura:altura+image_size_side, largura:largura+image_size_side] |
|
|
|
|
|
resized = cv2.resize(crop_img, tam_salvar, interpolation = cv2.INTER_AREA) |
|
resized_mask = cv2.resize(crop_mask, tam_salvar, interpolation = cv2.INTER_AREA) |
|
|
|
|
|
if (crop_polygon.intersects(frame_polygon)): |
|
area_interseccao = crop_polygon.intersection(frame_polygon).area |
|
area_crop_polygon = crop_polygon.area |
|
|
|
if(area_interseccao/area_crop_polygon > percentual_minimo and area_interseccao/area_crop_polygon <= percentual_maximo): |
|
cv2.imwrite(os.path.join(path_to_save, row['set'], 'image', row['class']+'-'+ row['name'].split('.')[0]+'-{}-{}-{}-{},{}_{},{}.jpg'.format(n_frame, c, k, largura, altura, largura+image_size_side, altura+image_size_side)), resized) |
|
cv2.imwrite(os.path.join(path_to_save, row['set'], 'mask', row['class']+'-'+ row['name'].split('.')[0]+'-{}-{}-{}-{},{}_{},{}.png'.format(n_frame, c, k, largura, altura, largura+image_size_side, altura+image_size_side)), resized_mask) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
largura = largura + stride |
|
k = k+1 |
|
|
|
altura = altura+stride |
|
|
|
except: |
|
print("Something else went wrong") |
|
else: |
|
continua = False |
|
print('read falso') |
|
|
|
|
|
for i in range(next_frame): |
|
cap.read() |
|
|
|
|
|
n_frame = n_frame + next_frame + 1 |
|
|
|
|
|
if n_frame > length-10: |
|
continua = False |
|
|
|
|
|
|
|
|
|
cap.release() |