def __init__(self,
label,
classes=None,
augmentation=None,
preprocessing=None):
self.gluoncv_dataset = CitySegmentation(split=label)
self.images_fps = self.gluoncv_dataset.images
self.masks_fps = self.gluoncv_dataset.mask_paths
# convert str names to class values on masks
self.class_values = [
self.CLASSES.index(cls.lower()) for cls in classes
]
self.augmentation = augmentation
self.preprocessing = preprocessing
import numpy as np
import tensorflow as tf
import keras
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, ReLU, BatchNormalization, Softmax, UpSampling2D
from gluoncv.data import CitySegmentation
train_dataset = CitySegmentation(split='train')
val_dataset = CitySegmentation(split='val')
train_examples = len(train_dataset)
val_examples = len(val_dataset)
#####################
def model_add(model, layers):
for l in layers:
model.add(l)
#####################
'''
x_train, y_train = zip(*train_dataset)
x_val, y_val = zip(*val_dataset)
'''
'''
count = 0
x_train = []; y_train = []
for (x, y) in train_dataset:
print (count)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
import numpy as np
import mxnet as mx
import tensorflow as tf
import keras
import queue
import threading
from gluoncv.data import CitySegmentation
from lib.SegNet import SegNet
####################################
train_dataset = CitySegmentation(split='train')
train_examples = len(train_dataset)
val_dataset = CitySegmentation(split='val')
val_examples = len(val_dataset)
batch_size = 5
epochs = 10
train_dataset_np = train_dataset.asnumpy()
####################################
def fill_queue(d, q):
ii = 0