def main():
print ("Enter input query (of the form '[<E,t> <J,t>][M, A]' ): ")
#input_query = input()
#input_query = '[<A,f>][B, J]'
#input_query = '[<J,t> <E,f>][B, M]'
input_query = '[<M,t> <J,f>][B, E]'
print ('Inference by Enumeration: ')
net = AlarmNetwork()
random = psuedorandom()
exactInfer = Inference(net, 0, 0, random )
print ( exactInfer.infer(input_query)[0])
print ("Enter sample count: ")
#s = input()
s = 10000
NoOfSamples = int(s)
print ('Inference by prior sampling: ')
priorSamplingInfer = Inference(net, 1, NoOfSamples, random)
print (priorSamplingInfer.infer(input_query)[0])
print ('Inference by Rejection sampling: ')
rejectionSamplingInfer = Inference(net, 2, NoOfSamples, random)
print (rejectionSamplingInfer.infer(input_query)[0])
print ('Inference by likelihood sampling: ')
likelihoodInfer = Inference(net, 3, NoOfSamples, random)
print (likelihoodInfer.infer(input_query)[0])
bool_near = (mask == 3)
bool_mid = (mask == 2)
bool_far = (mask == 1)
cumulative_near = bool_near
cumulative_near_len = cumulative_near.sum()
cumulative_mid = np.logical_or(bool_near, bool_mid)
cumulative_mid_len = cumulative_mid.sum()
cumulative_far = np.logical_or(cumulative_mid, bool_far)
cumulative_far_len = cumulative_far.sum()
while True:
frame = cap.read()
seg = model.infer(frame)
seg = np.expand_dims(seg, axis=2)
near = float(np.logical_and(
(seg == 0), cumulative_near).sum()) / cumulative_near_len
mid = float(np.logical_and(
(seg == 0), cumulative_mid).sum()) / cumulative_mid_len
far = float(np.logical_and(
(seg == 0), cumulative_far).sum()) / cumulative_far_len
obsPub.publish(Float32MultiArray(data=[near, mid, far]))
alpha = 0.4
overlay = frame.copy()
overlay[np.where((seg > 0).all(axis=2))] = green
def main(unused_argv):
global config
if len(unused_argv) != 1: # prints a message if you've entered flags incorrectly
raise Exception("Problem with flags: %s" % unused_argv)
print('Starting %s in %s mode...' % (FLAGS.exp_name, FLAGS.mode))
FLAGS.log_root = os.path.join(FLAGS.log_root, FLAGS.exp_name)
if not os.path.exists(FLAGS.log_root):
if FLAGS.mode == "train":
os.makedirs(FLAGS.log_root)
else:
raise Exception("Logdir %s doesn't exist. Run in train mode to create it." % (FLAGS.log_root))
hparam_list = ['mode', 'lr', 'rand_unif_init_mag', 'trunc_norm_init_std', 'max_grad_norm','hidden_dim', 'emb_dim', 'batch_size', 'max_dec_steps','max_bac_enc_steps', 'max_con_enc_steps', 'max_span_len', 'multi_hop_span_pre_mode', 'matching_layer', 'matching_gate']
hps_dict = {}
for key, val in FLAGS.__flags.items():
if key in hparam_list:
hps_dict[key] = val
hps = namedtuple("HParams", hps_dict.keys())(**hps_dict)
vocab = Vocab(FLAGS.vocab_path, FLAGS.vocab_size)
tf.set_random_seed(111)
if hps.mode == 'train':
batcher = Batcher(FLAGS.train_path, vocab, hps, single_pass=False)
print("creating model...")
model = Model(hps, vocab)
train(model, batcher)
elif hps.mode == 'val':
train_dir = os.path.join(FLAGS.log_root, "train")
hps = hps._replace(batch_size=1)
infer_model_hps = hps._replace(max_dec_steps=1)
try:
r = open(os.path.join(train_dir, "finished_val_models.json"), 'r', encoding='utf-8')
finished_val_models = json.load(r)
r.close()
except FileNotFoundError:
finished_val_models = {"finished_val_models": []}
w = open(os.path.join(train_dir, "finished_val_models.json"), 'w', encoding='utf-8')
json.dump(finished_val_models, w)
w.close()
while True:
ckpt = tf.train.get_checkpoint_state(train_dir)
if ckpt and ckpt.model_checkpoint_path:
for ckpt_path in list(ckpt.all_model_checkpoint_paths):
if ckpt_path in finished_val_models["finished_val_models"]:
print("Val_mode: %s already has been evaluated, skip it" % ckpt_path)
pass
else:
print("Val_mode: start new eval %s" % ckpt_path)
batcher = Batcher(FLAGS.dev_path, vocab, hps, single_pass=True)
model = Model(infer_model_hps, vocab)
val_infer = Inference(model, batcher, vocab, ckpt_path)
val_infer.infer()
tf.reset_default_graph()
finished_val_models["finished_val_models"].append(ckpt_path)
w = open(os.path.join(train_dir, "finished_val_models.json"), 'w', encoding='utf-8')
json.dump(finished_val_models, w)
w.close()
print("Val_mode: finished one eval %s" % ckpt_path)
print("Val_mode: current iterations of all_model_checkpoint_paths are completed...")
print("Val_mode: finished %d modes" % len(finished_val_models["finished_val_models"]))
if len(finished_val_models["finished_val_models"]) == FLAGS.epoch_num:
print("All val is ended")
break
else:
print('Val_mode: wait train finish the first epoch...')
time.sleep(60)
elif hps.mode == 'test':
hps = hps._replace(batch_size=1)
batcher = Batcher(FLAGS.test_path, vocab, hps, single_pass=True)
infer_model_hps = hps._replace(max_dec_steps=1)
model = Model(infer_model_hps, vocab)
if FLAGS.test_model_dir is None:
raise Exception("should appoint the test_model_dir")
test_infer = Inference(model, batcher, vocab, FLAGS.test_model_dir)
test_infer.infer()
else:
raise ValueError("The 'mode' flag must be one of train/val/test")
