def test_single_h5(FLAGS, h5_weights_path):
if not os.path.isfile(h5_weights_path):
print('%s is not a h5 weights file path' % h5_weights_path)
return
optimizer = adam(lr=FLAGS.learning_rate, clipnorm=0.001)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
load_weights(model, FLAGS.eval_weights_path)
img_names, test_data, test_labels = load_test_data(FLAGS)
predictions = model.predict(test_data, verbose=0)
right_count = 0
error_infos = []
for index, pred in enumerate(predictions):
pred_label = np.argmax(pred, axis=0)
test_label = test_labels[index]
if pred_label == test_label:
right_count += 1
else:
error_infos.append('%s, %s, %s\n' % (img_names[index], test_label, pred_label))
accuracy = right_count / len(img_names)
print('accuracy: %s' % accuracy)
result_file_name = os.path.join(os.path.dirname(h5_weights_path),
'%s_accuracy.txt' % os.path.basename(h5_weights_path))
with open(result_file_name, 'w') as f:
f.write('# predict error files\n')
f.write('####################################\n')
f.write('file_name, true_label, pred_label\n')
f.writelines(error_infos)
f.write('####################################\n')
f.write('accuracy: %s\n' % accuracy)
print('end')
def load_weights_save_pb(FLAGS):
optimizer = adam(lr=FLAGS.learning_rate, clipnorm=0.001)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
load_weights(model, FLAGS.freeze_weights_file_path)
save_pb_model(FLAGS, model)
def load_weights_save_pb(FLAGS):
optimizer = Nadam(lr=FLAGS.learning_rate,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-08,
schedule_decay=0.004)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
load_weights(model, FLAGS.freeze_weights_file_path)
save_pb_model(FLAGS, model)
def test_single_h5(FLAGS, h5_weights_path):
if not os.path.isfile(h5_weights_path):
print('%s is not a h5 weights file path' % h5_weights_path)
return
optimizer = Nadam(lr=FLAGS.learning_rate,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-08,
schedule_decay=0.004)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
#model = model_fn_SE_ResNet50(FLAGS, objective, optimizer, metrics)
load_weights(model, FLAGS.eval_weights_path)
img_names, test_data, test_labels = load_test_data(FLAGS)
predictions = model.predict(test_data, verbose=0)
right_count = 0
error_infos = []
print("img len :")
print(len(img_names))
print("test_data len :")
print(len(test_data))
print("test_labels len :")
print(len(test_labels))
for index, pred in enumerate(predictions):
pred_label = np.argmax(pred, axis=0) + 1
test_label = test_labels[index]
if pred_label == test_label:
print("{},{},{}".format(img_names[index], test_label, pred_label))
right_count += 1
else:
error_infos.append('%s, %s, %s\n' %
(img_names[index], test_label, pred_label))
accuracy = right_count / len(img_names)
print('accuracy: %s' % accuracy)
result_file_name = os.path.join(
os.path.dirname(h5_weights_path),
'%s_accuracy.txt' % os.path.basename(h5_weights_path))
with open(result_file_name, 'w') as f:
f.write('# predict error files\n')
f.write('####################################\n')
f.write('file_name, true_label, pred_label\n')
f.writelines(error_infos)
f.write('####################################\n')
f.write('accuracy: %s\n' % accuracy)
print('end')
def test_batch_h5(FLAGS):
"""
test all the h5 weights files in the model_dir
"""
optimizer = adam(lr=FLAGS.learning_rate, clipnorm=0.001)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
img_names, test_data, test_labels = load_test_data(FLAGS)
file_paths = mox.file.glob(os.path.join(FLAGS.eval_weights_path, '*.h5'))
for h5_weights_path in file_paths:
model = load_weights(model, h5_weights_path)
predictions = model.predict(test_data, verbose=0)
right_count = 0
error_infos = []
for index, pred in enumerate(predictions):
pred_label = np.argmax(pred, axis=0)
test_label = test_labels[index]
if pred_label == test_label:
right_count += 1
else:
error_infos.append('%s, %s, %s\n' % (img_names[index], test_label, pred_label))
accuracy = float(right_count) / len(img_names)
print('accuracy: %s' % accuracy)
result_file_name = os.path.join(os.path.dirname(h5_weights_path),
'%s_accuracy.txt' % os.path.basename(h5_weights_path))
with mox.file.File(result_file_name, 'w') as f:
f.write('# predict error files\n')
f.write('####################################\n')
f.write('file_name, true_label, pred_label\n')
f.writelines(error_infos)
f.write('####################################\n')
f.write('accuracy: %s\n' % accuracy)
print('accuracy result file saved as %s' % result_file_name)
print('end')
def test_single_h5(FLAGS, h5_weights_path):
if not os.path.isfile(h5_weights_path):
print('%s is not a h5 weights file path' % h5_weights_path)
return
optimizer = Nadam(lr=FLAGS.learning_rate,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-08,
schedule_decay=0.004)
objective = 'categorical_crossentropy'
metrics = ['accuracy']
model = model_fn(FLAGS, objective, optimizer, metrics)
load_weights(model, FLAGS.eval_weights_path)
img_names, test_data = load_test_data(FLAGS)
predictions = model.predict(test_data, verbose=0)
test_labels = []
for index, pred in enumerate(predictions):
pred_label = np.argmax(pred, axis=0)
test_labels.append(pred_label + 1)
img_names = [x.split('/')[-1] for x in img_names]
df = pd.DataFrame({"FileName": img_names, "type": test_labels})
df.to_csv('result.csv', index=0)
parser.add_argument(
"--model_dir", type=str, default=getenv("SM_MODEL_DIR", "/opt/ml/models")
)
parser.add_argument(
"--data_dir",
type=str,
default=getenv("SM_CHANNEL_TRAINING", "/opt/ml/input/data/training"),
)
parser.add_argument(
"--output_dir", type=str, default=getenv("SM_OUTPUT_DIR", "/opt/ml/output/")
)
args, _ = parser.parse_known_args()
# call preprocessing on the data
df = preprocess(join(args.data_dir, "public.csv.gz"))
# load the model
model = model_fn(args.model_dir)
logger.info("creating predictions")
inputs = df.drop(columns=target_columns)
predictions = model.predict(inputs)
logger.info(f"predictions have shape of {predictions.shape}")
# save the predictions
pd.DataFrame(predictions).to_csv(
join(args.output_dir, "public.csv.out"), index=False, header=False
)