def main(): trains, chs = read_trains() n_net = NeuralNetwork() if os.path.exists(TRAINED_FPATH): n_net.load_trained_file(TRAINED_FPATH) else: n_net.set_trains(trains) train(n_net, DESIRED_ACCURACY) n_net.save_trained_file(TRAINED_FPATH) test_inputs(n_net, chs)
class MainWnd(QWidget):
def __init__(self, parent=None):
super(MainWnd, self).__init__(parent)
self.grid = Grid(self)
self.grid.move(10, 10)
self.next_g = QPushButton("&Next", self)
self.next_g.clicked.connect(self.on_next)
self.next_g.move(self.grid.width() + SPACING * 2, SPACING)
self.next_g.resize(PANEL_ITEM_W, PANEL_ITEM_H)
self.next_g.show()
self.prev_g = QPushButton("&Previous", self)
self.prev_g.clicked.connect(self.on_prev)
self.prev_g.move(self.grid.width() + SPACING * 2, SPACING * 2 + PANEL_ITEM_H)
self.prev_g.resize(PANEL_ITEM_W, PANEL_ITEM_H)
self.prev_g.show()
self.clear_g = QPushButton("&Clear", self)
self.clear_g.clicked.connect(self.on_clear)
self.clear_g.move(self.grid.width() + SPACING * 2, SPACING * 3 + PANEL_ITEM_H * 2)
self.clear_g.resize(PANEL_ITEM_W, PANEL_ITEM_H)
self.clear_g.show()
self.train_g = QPushButton("&Train", self)
self.train_g.clicked.connect(self.on_train)
self.train_g.move(self.grid.width() + SPACING * 2, SPACING * 4 + PANEL_ITEM_H * 3)
self.train_g.resize(PANEL_ITEM_W, PANEL_ITEM_H)
self.train_g.show()
self.out_g = QLabel(self)
self.out_g.setAlignment(Qt.AlignCenter)
self.out_g.setFont(QFont(None, 30))
self.out_g.move(self.grid.width() + SPACING * 2, SPACING * 5 + PANEL_ITEM_H * 4)
self.out_g.resize(PANEL_ITEM_W, PANEL_ITEM_H)
self.out_g.show()
try:
self.trains, self.chs = read_trains()
except IOError:
self.trains, self.chs = [], []
self.n_net = NeuralNetwork()
try:
self.n_net.load_trained_file(TRAINED_FPATH)
except IOError:
pass
try:
self.tests = read_tests()
random.shuffle(self.tests)
except IOError:
self.tests = []
self.test_idx = -1
self.cur_test = None
if self.tests:
self.set_test_idx(0)
self.train_done.connect(self.on_train_done)
self.setWindowTitle("Optical Character Recognition")
self.resize(self.grid.width() + SPACING * 3 + PANEL_ITEM_W, self.grid.height() + SPACING * 2)
self.show()
def set_test_idx(self, idx):
if idx < 0 or idx >= len(self.tests):
raise IndexError, "out of range"
self.cur_test = self.tests[idx]
self.test_idx = idx
self.grid.load_vec(self.cur_test[1])
self.test_grid()
def on_next(self):
try:
self.set_test_idx(self.test_idx + 1)
except IndexError:
QMessageBox.warning(self, None, "No more tests.")
def on_prev(self):
try:
self.set_test_idx(self.test_idx - 1)
except IndexError:
QMessageBox.warning(self, None, "No more tests.")
def on_clear(self):
self.grid.clear()
self.test_grid()
def on_train(self):
QMessageBox.warning(self, None, "Training sequence started. May take several minitues to finish.")
def proc():
self.n_net.set_trains(self.trains)
train(self.n_net, DESIRED_ACCURACY)
self.n_net.save_trained_file(TRAINED_FPATH)
self.train_done.emit()
thread.start_new_thread(proc, ())
@Slot()
def on_train_done(self):
QMessageBox.warning(self, None, "Training done.")
train_done = Signal()
def test_grid(self):
try:
ch = test_input(self.n_net, self.grid.get_vec(), self.chs)
except ValueError:
ch = "?"
self.out_g.setText(ch)
