class ConstructGrid:
def __init__(self, cellarray, modeltype):
self.cellarray = cellarray
self.recognizer = DigitRecognizer(modeltype)
self.finalgrid = [[0 for i in range(9)] for j in range(9)]
self.imagewritten = False
def constructgrid(self):
threshold = 5*255
for i in range(9):
for j in range(9):
#First preprocess and clean the board cell
tmp = np.copy(self.cellarray[i][j])
tmp = self.recognizer.preprocess_image(tmp)
cv2.imwrite("CleanedBoardCells/cell"+str(i)+str(j)+".jpg", tmp)
#Checking if the board cell is empty or not
finsum = 0
for k in range(28):
rowsum = sum(tmp[k])
finsum += rowsum
if finsum < threshold:
self.finalgrid[i][j] = 0
continue
if not self.imagewritten:
try:
os.remove("StagesImages/13.jpg")
os.remove("StagesImages/14.jpg")
except:
pass
cv2.imwrite("StagesImages/13.jpg", self.cellarray[i][j])
cv2.imwrite("StagesImages/14.jpg", tmp)
self.imagewritten = True
pred = self.recognizer.make_prediction(str("CleanedBoardCells/cell"+str(i)+str(j)+".jpg"))
self.finalgrid[i][j] = int(pred)
return self.finalgrid
class ConstructGrid:
def __init__(self, cellarray, modeltype):
self.cellarray = cellarray
self.recognizer = DigitRecognizer(modeltype)
self.finalgrid = [[0 for i in range(9)] for j in range(9)]
self.imagewritten = False
'''This function uses a threshold of 5 white pixels to determine if a grid cell is empty.
If the grid cell isn't empty, the prediction of the KNN is used to determine the number.
This function writes the 13th stage image to StagesImages and
finally returns the final grid of predicted numbers'''
def constructgrid(self):
threshold = 5 * 255
for i in range(9):
for j in range(9):
#First preprocess and clean the board cell
tmp = np.copy(self.cellarray[i][j])
tmp = self.recognizer.preprocess_image(tmp)
cv2.imwrite(
"CleanedBoardCells/cell" + str(i) + str(j) + ".jpg", tmp)
#Checking if the board cell is empty or not
finsum = 0
for k in range(28):
rowsum = sum(tmp[k])
finsum += rowsum
if finsum < threshold:
self.finalgrid[i][j] = 0
continue
if not self.imagewritten:
try:
os.remove("StagesImages/13.jpg")
os.remove("StagesImages/14.jpg")
except:
pass
cv2.imwrite("StagesImages/13.jpg", self.cellarray[i][j])
cv2.imwrite("StagesImages/14.jpg", tmp)
self.imagewritten = True
pred = self.recognizer.make_prediction(
str("CleanedBoardCells/cell" + str(i) + str(j) + ".jpg"))
self.finalgrid[i][j] = int(pred)
return self.finalgrid
def __init__(self, cellarray, modeltype):
self.cellarray = cellarray
self.recognizer = DigitRecognizer(modeltype)
self.finalgrid = [[0 for i in range(9)] for j in range(9)]
self.imagewritten = False
def train_start(self):
self.recognizer = DigitRecognizer()
self.recognizer.learn()
class MainWindow(QMainWindow):
def __init__(self):
super(MainWindow, self).__init__()
self.ui = loadUi(os.path.join(THIS_DIR, 'mainwindow.ui'), self)
self.thread = QThread()
try:
self.camera = CameraDevice()
except ValueError:
self.ui.video.setText("Device not found!\n\nIs FFMPEG available?")
else:
self.camera.frame_ready.connect(self.update_video_label)
self.ui.video.setMinimumSize(640 * 2, 480)
self.camera.moveToThread(self.thread)
self.ui.t_max.setValue(0)
self.ui.t_min.setValue(255)
self.ui.s_max.setValue(200)
self.ui.s_min.setValue(3)
self.update_values()
@pyqtSlot()
def start_recog(self):
if self.recognizer is not None:
self.camera.set_recog(self.recognizer)
else:
print("plz do load recogi")
@pyqtSlot()
def stop_recog(self):
self.camera.remove_recog()
@pyqtSlot()
def train_start(self):
self.recognizer = DigitRecognizer()
self.recognizer.learn()
@pyqtSlot()
def make_data(self):
print("Not supported.")
@pyqtSlot(int)
def t_max_changed(self, val):
if val < self.ui.t_min.value():
self.ui.t_min.setValue(val)
self.update_values()
@pyqtSlot(int)
def t_min_changed(self, val):
if val > self.ui.t_max.value():
self.ui.t_max.setValue(val)
self.update_values()
@pyqtSlot(int)
def s_max_changed(self, val):
if val < self.ui.s_min.value():
self.ui.s_min.setValue(val)
self.update_values()
@pyqtSlot(int)
def s_min_changed(self, val):
if val > self.ui.s_max.value():
self.ui.s_max.setValue(val)
self.update_values()
@pyqtSlot(QImage)
def update_video_label(self, image):
pixmap = QPixmap.fromImage(image)
self.ui.video.setPixmap(pixmap)
self.ui.video.update()
def update_values(self):
self.camera.set_values(self.ui.t_max.value(), self.ui.t_min.value(),
self.ui.s_max.value(), self.ui.s_min.value())