config.py

H_size = 256
W_size = 256
batch_size = 10
epochs = 1000000
tasks = 18
processes = 18
inter_op_parallelism_threads=7
allow_growth = True
noise_size = 100
cond_emb_dim = 100
input_channels = 3
D_step = 1000
save_step = 50000
load_previous_weigth = False
reuse_dataset = True
max_pic_save = 50
max_images_for_activity = 20
load_pretrained_inception = True
max_no_D_impro = 2
max_no_G_impro = 20
D_loss_thresh = 0.1
Ng = 32
Nd = 64
lambda_1 = 1
lambda_2 = 5

alfa_color = 50
alfa_classification_D = 10
alfa_classification_G = 0
alfa_mean = 0

action = 'karaoke'
visible_gpu = '2,3'
#action = 'moving_child'
#visible_gpu = '0,1'
dataset_folder = '/media/Data/Lorenzo/Code/Kinetcs_no_vid/processed_frames/' + action + '/all_conditioning/ClassesIcons'
ckpt_load_folder = './checkpoint/'  + action + '/'
model_filename = ckpt_load_folder + 'Net_weigths.model'
checkpoint_file_check = ckpt_load_folder +  '/checkpoint'
out_path = './results/' + action + '/'
mean_size = 32
norm_min = 0
norm_max = 1

num_patches = 8