Upload 6 files

scripts/bboxAnnotator.py

@@ -0,0 +1,27 @@
+import numpy as np
+import cv2
+
+class BBOXAnnotator:
+    def draw_bbox(self, ImgPath, bboxPlate, plateNumber):
+        img = ImgPath
+
+        for idc, bb in enumerate(bboxPlate):
+            # print('bboxp : ', bboxp)
+            # print('bb : ', bb)
+            x1, y1, x2, y2 = map(int, bb)  # Ubah koordinat ke dalam integer
+            color = (0, 255, 0)  # Warna bounding box (hijau)
+            thickness = 2  # Ketebalan garis bounding box
+
+            # Gambar bounding box pada gambar
+            cv2.rectangle(img, (x1, y1), (x2, y2), color, thickness)
+
+            # Tambahkan label nama di atas bounding box
+            if len(plateNumber) == len(bboxPlate):
+                print('plateNumber : ', plateNumber)
+                # print('idc : ', idc)
+                cv2.putText(img, plateNumber[idc], (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, thickness)
+            
+        if len(plateNumber) == 0:
+            plateNumber = ['No Plate Detected']
+        
+        return img, plateNumber

scripts/charExtraction.py

@@ -0,0 +1,79 @@
+from scripts.extractor import CharExtractor
+from ultralytics import YOLO
+import numpy as np
+import cv2
+
+
+class CharExtraction:
+    def __init__(self, wPlatePath, wCharacterPath, classList, sizePlate, conf):
+        self.wPlatePath = wPlatePath
+        self.wCharacterPath = wCharacterPath
+        self.classList = classList
+        self.width, self.height = sizePlate
+        self.conf = conf
+        
+        self.model_plat = YOLO(self.wPlatePath)
+        self.model_char = YOLO(self.wCharacterPath)
+        
+        self.extractor = CharExtractor()
+        
+    def predict(self, image, conf):
+        if conf == None:
+            self.conf = 0.3
+        else:
+            self.conf = conf
+
+        charpredict = []
+        platePrediction = self.model_plat(image, conf=self.conf, verbose=False) ##Melakukan prediksi dan rectange gambar
+        platBoxes, croppedPlateImg, confident = self._get_cropped_plate(platePrediction)
+        # print('lcrop : ', len(croppedPlateImg))
+        
+        for croppImg in croppedPlateImg:
+            characterPrediction = self.model_char(croppImg, conf=self.conf, verbose=False)
+            charpredict.append(characterPrediction)
+            
+        # print('lchar : ', len(charpredict))
+        characterBBOX, characterClass = self._get_bboxClass_char(charpredict, self.classList)
+        # print('charBox : ', len(characterBBOX))
+        plateNumber = self.extractor._extract_bbox(characterBBOX, characterClass)
+        # print('Plate Number : ', plateNumber)   
+        
+        # print('plateboxes : ', platBoxes)
+        # print('platenum : ', plateNumber)        
+        return platBoxes, plateNumber, confident             
+    
+    def _get_cropped_plate(self, imgPlate): # Get cropped plate image from whole image
+        croppedImage = [] ##Bisa berisi banyak bbox
+        boxes = imgPlate[0].boxes
+        originalImage = imgPlate[0].orig_img
+        confid = boxes.conf.numpy()
+        h,w,c = originalImage.shape
+        if len(boxes) != 0:
+            for box in boxes.xyxy:
+                x1,y1,x2,y2 = int(box[0]), int(box[1]), int(box[2]), int(box[3])
+                croppedPlateImage = originalImage[y1:y2, x1:x2, :]
+                croppedPlateImage = cv2.resize(croppedPlateImage, (self.width, self.height))
+                croppedImage.append(croppedPlateImage)
+        else:
+            croppedImage.append(originalImage)
+        
+        return boxes.xyxy, croppedImage, confid
+    
+    def _get_bboxClass_char(self, imgCroppedPlate, classList):
+        boxChar, classC = [], []
+        for imgCropped in imgCroppedPlate:
+            boxes = imgCropped[0].boxes
+            if len(boxes.cls) != 0:
+                # print(boxes.cls)
+                classId = [int(x) for x in boxes.cls.numpy()]
+                # print('classId : ', classId)
+                classCharacter = np.array(classList[classId])
+                characterBBOX = boxes.xywh
+                boxChar.append(characterBBOX)
+                classC.append(classCharacter)
+            # else:
+            #     classCharacter = []
+            #     characterBBOX = []
+            
+        return boxChar, classC
+        

scripts/extractor.py

@@ -0,0 +1,76 @@
+import cv2
+import numpy as np
+import matplotlib.pyplot as plt
+
+class CharExtractor:
+    def __init__(self):
+        self.classes = None
+        self.class_char = []
+        self.y_class = None
+        self.x_class = None
+        self.bbox = None
+        self.x_bbox = None
+        self.y_bbox = None
+    
+    def _extract_bbox(self, bboxes, classes):
+        plate_number = []
+        self.bbox = bboxes
+        self.classes = classes
+        
+        for ix,bbox in enumerate(bboxes):
+            if len(bbox) != 0:
+                lowest_x, highest_y = self._get_xy(bbox, ix)
+                print('x_class : ', self.x_class)
+                # print('y_bbox : ', self.y_bbox)
+                for idx, point in enumerate(self.y_bbox):
+                    x, y = point[0], point[1]
+                    if y <= highest_y and x >= lowest_x and self.x_class.size > 1:
+                        self.class_char.append(self.x_class[idx])
+
+                plate_num = ''.join(str(v) for v in self.class_char)
+                self.class_char.clear()
+        
+            else:
+                plate_num = ''
+            
+            plate_number.append(plate_num)
+        
+        return plate_number
+    
+    def _get_xy(self, bbox, ix):
+        y_id = bbox[:, 1].argsort()
+        # print('y_id : ', y_)
+        if len(bbox[y_id]) >= 4:
+            sorted_y = bbox[y_id][:-4]
+            self.y_bbox = sorted_y
+            # print('y_class : ', self.classes[ix])
+            # print('yid : ', y_id)
+            self.y_class = np.array(self.classes[ix][y_id])[:-4]
+            highest_y = max([x[1] for x in sorted_y])
+            
+            x_id = sorted_y[:,0].argsort()
+            sorted_x = sorted_y[x_id]
+            self.x_bbox = sorted_x
+            # print('char : ', self.y_class)
+            self.x_class = self.y_class[x_id]
+            lowest_x = min([x[0] for x in sorted_x])
+            
+        else:
+            sorted_y = bbox[y_id]
+            self.y_bbox = sorted_y
+            # print('y_class : ', self.classes)
+            # print('yid : ', y_id)
+            self.y_class = np.array(self.classes[ix][y_id])
+            highest_y = max([x[1] for x in sorted_y])
+            
+            x_id = sorted_y[:,0].argsort()
+            sorted_x = sorted_y[x_id]
+            self.x_bbox = sorted_x
+            # print('char : ', self.y_class)
+            self.x_class = self.y_class
+            lowest_x = min([x[0] for x in sorted_x])
+        
+        return lowest_x, highest_y
+ 
+            
+