import numpy as np import sys import pickle import random from PIL import Image import cv2 from keras.utils import np_utils import config from sklearn.metrics import classification_report from keras import backend as K import math server = config.server() data_output_path = config.data_output_path() data_folder_rgb = r'{}rgb/'.format(data_output_path) data_folder_seq = r'{}seq3/'.format(data_output_path) def getTrainData(keys, batch_size, classes, mode, train, opt_size, seq=False): """ mode 1: Single Stream mode 2: Two Stream mode 3: Multiple Stream """ while 1: for i in range(0, len(keys), batch_size): if not seq: if mode == 1: X_train, Y_train = stack_single_stream( chunk=keys[i:i + batch_size], opt_size=opt_size,
old_epochs=0, cross_index=cross_index) model.load_weights('weights-old/save-imp/{}_{}e_cr{}.h5'.format( pre_file, 45, cross_index)) data_type = [0] else: print "Error stream" result_model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy']) out_file = r'{}database/{}-test{}-split{}.pickle'.format( config.data_output_path(), dataset, seq_len, cross_index) with open(out_file, 'rb') as f2: keys = pickle.load(f2) if index > (len(keys) - 1): print 'Out of number data test' sys.exit() class_file = 'data/{}-classInd.txt'.format(dataset) classInd = [] with open(class_file) as f0: for line in f0: class_name = line.rstrip() if class_name: classInd.append(class_name)