# 6 = Charizard chosen_pokemon_2 = 9 # Squirtle # chosen_pokemon_2 = 13 # Caterpie chosen_pokemon = 30 # Pikachu chosen_pokemon = 146 # Jolteon # chosen_pokemon = 161 # Dragonite # chosen_pokemon_2 = 252 # Porygon 2 # chosen_pokemon_2 = 330 # Aaron # chosen_pokemon = 427 # Jirachi # chosen_pokemon_2 = 622 # Darmanitan # chosen_pokemon = 972 # shield doggo. interpolation_intervals = 7 X_full_HSV, Y_full_HSV = \ utilities.prepare_dataset_for_input_layer("pokedataset32_full_HSV_Two_Hot_Encoded.h5") Y_full_HSV = Y_full_HSV * 0.5 expanded_full_X_HSV = np.append(X_full_HSV, Y_full_HSV, axis=1) print("getting network to load model*******************") network_instance = utilities.get_network() network_instance = tflearn.regression( network_instance, optimizer='adam', metric='R2', loss=utilities.vae_loss, learning_rate=0.00001) # adagrad? #adadelta #nesterov did good, model = tflearn.DNN(network_instance) print("LOADING MODEL.")
import tflearn import h5py import pokedataset32_vae_functions as utilities from PIL import Image import colorsys # current_dataset = 'pokedataset' current_dataset = 'anime_faces_' use_anime_with_types = True if not use_anime_with_types: X_full_HSV, Y_full_HSV, X_full_RGB, Y_full_RGB, X, Y, test_X, test_Y = utilities.ready_all_data_sets( current_dataset) else: X, Y = utilities.prepare_dataset_for_input_layer( 'anime_faces_32_train_HSV_Two_Hot_Encoded_Augmented_With_Types.h5', in_dataset_x_label='anime_faces_32_X', in_dataset_y_label='anime_faces_32_Y') test_X, test_Y = utilities.prepare_dataset_for_input_layer( 'anime_faces_32_train_HSV_Two_Hot_Encoded_Augmented_With_Types.h5', in_dataset_x_label='anime_faces_32_X_test', in_dataset_y_label='anime_faces_32_Y_test') X_full_RGB, Y_full_RGB = utilities.prepare_dataset_for_input_layer( 'anime_faces_32_full_RGB_Two_Hot_Encoded.h5', in_dataset_x_label='anime_faces_32_X', in_dataset_y_label='anime_faces_32_Y') X_first_half = X[0:int(len(X) / 2)] Y_first_half = Y[0:int(len(Y) / 2)] test_X_first_half = test_X[0:int(len(test_X) / 2)] test_Y_first_half = test_Y[0:int(len(test_Y) / 2)] """X_second_half = X[int(len(X) / 2):]
import h5py import pokedataset32_vae_functions as utilities from PIL import Image import colorsys import math # current_dataset = 'pokedataset' # current_dataset = 'anime_faces_' # We don't need all of those. # X_full_HSV, Y_full_HSV, X_full_RGB, Y_full_RGB, X, Y, test_X, test_Y = utilities.ready_all_data_sets(current_dataset) # X_full_HSV, Y_full_HSV = utilities.prepare_dataset_for_input_layer('pokedataset32_full_HSV_Two_Hot_Encoded.h5') X_full_HSV, Y_full_HSV = utilities.prepare_dataset_for_input_layer( 'pokedataset32_full_HSV_Two_Hot_Encoded.h5', in_dataset_x_label='pokedataset32_X', in_dataset_y_label='pokedataset32_Y') # """ X_full_HSV_faces, Y_full_HSV_faces = utilities.prepare_dataset_for_input_layer( 'anime_faces_32_train_HSV_Two_Hot_Encoded_Augmented.h5', in_dataset_x_label='anime_faces_32_X', in_dataset_y_label='anime_faces_32_Y') X_test_HSV_faces, Y_test_HSV_faces = utilities.prepare_dataset_for_input_layer( 'anime_faces_32_train_HSV_Two_Hot_Encoded_Augmented.h5', in_dataset_x_label='anime_faces_32_X_test', in_dataset_y_label='anime_faces_32_Y_test') # """ """ # FOR DEBUGGING PURPOSES:
import numpy as np import tensorflow as tf import tflearn import matplotlib.colors import pokedataset32_vae_functions as utilities current_dataset = 'pokedataset' # current_dataset = 'anime_faces_' # X and Y are not used in this file. X_full_HSV, Y_full_HSV, X_full_RGB, Y_full_RGB, X, Y, test_X, test_Y = utilities.ready_all_data_sets( current_dataset) X_full_HSV_regional, Y_full_HSV_regional = \ utilities.prepare_dataset_for_input_layer("pokedataset32_full_HSV_Two_Hot_Encoded_Regional.h5") Y_full_HSV_regional = Y_full_HSV_regional * 0.5 expanded_full_X_HSV_regional = np.append(X_full_HSV_regional, Y_full_HSV_regional, axis=1) X_full_RGB_regional, Y_full_RGB_regional = \ utilities.prepare_dataset_for_input_layer("pokedataset32_full_RGB_Two_Hot_Encoded_Regional.h5") Y_full_RGB_regional = Y_full_HSV_regional * 0.5 expanded_full_X_RGB_regional = np.append(X_full_HSV_regional, Y_full_HSV_regional, axis=1) Y_full_HSV = Y_full_HSV * 0.50 small_X = np.concatenate((X[0:200], test_X[0:200]), axis=0) small_Y = np.concatenate((Y[0:200], test_Y[0:200]), axis=0) len_X_div_2 = int(len(X) / 2)
from __future__ import division, print_function, absolute_import import numpy as np import tensorflow as tf import tflearn import matplotlib.pyplot as plt import matplotlib.colors import pokedataset32_vae_functions as utilities X_full_HSV_Type_Swapped, Y_full_HSV_Type_Swapped = \ utilities.prepare_dataset_for_input_layer('pokedataset32_full_HSV_Two_Hot_Encoded_Type_Swapped.h5') X_full_RGB_Type_Swapped, Y_full_RGB_Type_Swapped = \ utilities.prepare_dataset_for_input_layer('pokedataset32_full_RGB_Two_Hot_Encoded_Type_Swapped.h5') Y_full_HSV_Type_Swapped = np.reshape(np.asarray(Y_full_HSV_Type_Swapped), newshape=[Y_full_HSV_Type_Swapped.shape[0], utilities.pokemon_types_dim]) expanded_full_X_HSV_Type_Swapped = np.append(X_full_HSV_Type_Swapped, Y_full_HSV_Type_Swapped, axis=1) print("getting network to load model*******************") network_instance = utilities.get_network() network_instance = tflearn.regression(network_instance, optimizer='adam', # optimizer='rmsprop', metric='R2', # loss='mean_square', loss=utilities.vae_loss, learning_rate=0.001) # adagrad? #adadelta #nesterov did good,