netfilename = 'net_python_cloud_detector4.mat' data = cloud.load_net(netfilename) net = data['net'] sample_mean = np.transpose(data['sample_mean']) sample_std = np.transpose(data['sample_std']) patchsize = data['patchsize'][0][0] offset = (patchsize - 1) / 2 inputsize = 7 * patchsize**2 #layer = old_net.layer; #layer.insert(0,nnet.layer(inputsize)) #net = nnet.net(layer); d = cloud.load_all_data('/local_scratch/cloudmasks/mat_training_notime') confusion_matrix = {} total_confusion_matrix = np.zeros((num_classes, num_classes)) test_confusion_matrix = np.zeros((num_classes, num_classes)) f_out = open(os.path.join(imgdir, 'classification_printout.txt'), 'w') np.set_printoptions(suppress=True) print('test rate for this dataset: ' + str(data['test_rate'])) for i in range(len(d[0])): A = d[0][i] MASK = d[1][i] MASKF = d[2][i] filename = d[3][i] [fname, fext] = os.path.splitext(filename)
import sys import numpy as np import gc import time use_sample_storage = True #off by default - below import may turn on from cloud_params import * from nnet_toolkit import nnet np.random.seed = randomseed1 inputsize = 7 * patchsize**2 offset = (patchsize - 1) / 2 d = cloud.load_all_data(data_path) estimated_load_percentage = np.min(class_load_percentage) estimated_size = int(1000.0 * 1000.0 * 44.0 * 2.4 * estimated_load_percentage) sample_list = np.zeros((estimated_size, inputsize), dtype=np.float32) class_list = np.zeros((estimated_size, 3), dtype=np.int8) #class_list = [] sample_list_test = [] class_list_test = [] samples_stored = 0 print('estimated size: ' + str(estimated_size)) for i in range(len(d[0])): A = d[0][0] MASK = d[1][0]
netfilename = 'net_python_cloud_detector4.mat' data = cloud.load_net(netfilename) net = data['net'] sample_mean = np.transpose(data['sample_mean']) sample_std = np.transpose(data['sample_std']) patchsize = data['patchsize'][0][0] offset = (patchsize-1)/2; inputsize = 7*patchsize**2 #layer = old_net.layer; #layer.insert(0,nnet.layer(inputsize)) #net = nnet.net(layer); d = cloud.load_all_data('/local_scratch/cloudmasks/mat_training_notime') confusion_matrix = {} total_confusion_matrix = np.zeros((num_classes,num_classes)) test_confusion_matrix = np.zeros((num_classes,num_classes)) f_out = open(os.path.join(imgdir,'classification_printout.txt'),'w') np.set_printoptions(suppress=True) print('test rate for this dataset: ' + str(data['test_rate'])) for i in range(len(d[0])): A = d[0][i] MASK = d[1][i] MASKF = d[2][i] filename = d[3][i] [fname, fext] = os.path.splitext(filename)
import sys import numpy as np import gc import time use_sample_storage = True #off by default - below import may turn on from cloud_params import * from nnet_toolkit import nnet np.random.seed = randomseed1; inputsize = 7*patchsize**2 offset = (patchsize-1)/2; d = cloud.load_all_data(data_path) estimated_load_percentage = np.min(class_load_percentage) estimated_size = int(1000.0*1000.0*44.0*2.4*estimated_load_percentage) sample_list = np.zeros((estimated_size,inputsize),dtype=np.float32) class_list = np.zeros((estimated_size,3),dtype=np.int8) #class_list = [] sample_list_test = [] class_list_test = [] samples_stored = 0 print('estimated size: ' + str(estimated_size)) for i in range(len(d[0])): A = d[0][0] MASK = d[1][0]