def train_model(model,
save_path=save_path,
learning_rate=0.01,
momentum=0.9,
loss="binary_crossentropy",
path_train=path_train):
train_dataset = dataset.CatDataset(path_train, im_width, im_height)
# train_dataset.augment()
X_train, y_train = train_dataset.X, train_dataset.Y
model.compile(optimizer=SGD(learning_rate=learning_rate,
momentum=momentum),
loss=loss,
metrics=["accuracy"])
callbacks = [
EarlyStopping(patience=10, verbose=1),
ReduceLROnPlateau(factor=0.1, patience=3, min_lr=0.00001, verbose=1),
ModelCheckpoint(save_path,
monitor='accuracy',
mode='max',
verbose=1,
save_best_only=True,
save_weights_only=True)
]
results = model.fit(X_train,
y_train,
batch_size=32,
epochs=100,
callbacks=callbacks)
return results
def test_model(weight_path, threshold, path_test=path_test):
if not os.path.exists("./Output"):
os.mkdir("./Output")
if not os.path.exists("./Output/x"):
os.mkdir("./Output/x")
if not os.path.exists("./Output/y"):
os.mkdir("./Output/y")
if not os.path.exists("./Output/pred"):
os.mkdir("./Output/pred")
test_dataset = dataset.CatDataset(path_test, im_width, im_height)
X_test, y_test = test_dataset.X, test_dataset.Y
model.load_weights(weight_path)
pred_test = model.predict(X_test, verbose=1)
# dice_coe_result = dice_coe(pred_test, y_test, loss_type="sorensen")
# print('Sorensen Dice Coefficient result is {}'.format(dice_coe_result))
for i in range(pred_test.shape[0]):
test_pred = pred_test[i]
test_y = y_test[i] * 255
test_x = X_test[i] * 255
test_pred[test_pred > threshold] = 255
test_pred[test_pred <= threshold] = 0
save_img("./Output/x/x_{}.jpg".format(i), test_x)
save_img("./Output/y/y_{}.jpg".format(i), test_y)
save_img("./Output/pred/pred_{}.jpg".format(i), test_pred)
model.compile(optimizer=SGD(),
loss="binary_crossentropy",
metrics=["accuracy"])
model.evaluate(X_test, y_test)
def predict(weight_path, pred_path, threshold=0.2, out_path="./Output"):
pred_dataset = dataset.CatDataset(pred_path, im_width, im_height)
X_pred, y_pred = pred_dataset.X, pred_dataset.Y
model.load_weights(weight_path)
preds = model.predict(X_pred, verbose=1)
for i in range(preds.shape[0]):
pred = preds[i]
print(np.where(pred > threshold))
pred[pred > threshold] = 255.0
pred[pred < threshold] = 0
save_img(f'{out_path}/{i}.jpg', pred)