merge Anpr into main code

api/OCR_model.py

+from paddleocr import PaddleOCR, draw_ocr
+from PIL import Image
+from PIL import ImageFont
+import re
+
+
+import os
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+# Paddleocr supports Chinese, English, French, German, Korean, and Japanese.
+# You can set the parameter `lang` as `ch`, `en`, `fr`, `german`, `korean`, `japan`
+# to switch the language model in order.
+ocr = PaddleOCR(use_angle_cls=True, lang='en')  # need to run only once to download and load the model into memory
+provineces = ['NP' , 'SP' , 'EP' , 'WP', 'SG' , 'NC' , 'NW' , 'CP' , 'UP'  ]
+numbers_range = '0123456789'
+letters_range = 'W,E,S,C,N,P,U'
+def1 = 'PF'
+def2 = '2969'
+def get_licsence_number( model , image_path):
+
+
+    result = model.ocr(image_path, cls=True)
+    result = ocr.ocr(image_path, cls=True)
+    res = result[0][0][1][0]
+
+    return res
+     
+
+def pattern(default , number):
+
+    
+
+    if number is not None :
+        letters = number.split('-')[0].strip()
+        if len(letters) > 2 and letters.isupper():
+            province = letters[:len(letters) - 2]
+
+            matches = [item for item in provineces if item.startswith(province)]
+            if len(matches) > 0:
+                p = matches[0]
+                result = p + number[len(letters) - 2:] 
+                return result
+            else:
+                return None
+   
+
+
+
+

api/app.py

@@ -4,14 +4,25 @@ import torch
 import hashlib
 import os
 import threading
+import random
 import math
+from ultralytics import YOLO
+from paddleocr import PaddleOCR
+from OCR_model import get_licsence_number , pattern
+import OCR_model as OM 
+from sending_mails import sending
+
+email = ""
+password = ""
+to = ""
 
-app = Flask(__name__)
 
+app = Flask(__name__)
+ocr = PaddleOCR(use_angle_cls=True, lang='en') 
 # Load your custom-trained YOLOv5 model with your weights
 # Update the absolute path to your custom YOLOv5 model code directory
 yolov5_code_directory = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\yolov5'
-
+count = 0
 # Load your custom-trained YOLOv5 model with your weights
 weights_path = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\best_with_200_epochs.pt'
 weights_path_speed = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\best_speed_weights.pt'
@@ -26,8 +37,8 @@ class_mapping_speed = {"lorry": 0, "car": 1, "bus": 2, "van": 3, "truck": 4, "do
 vehicle_labels_speed = [label for label, index in sorted(class_mapping_speed.items(), key=lambda x: x[1])]
 
 # Video input path
-video_path = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\test1 (1) compressed.mp4'
-video_path_speed = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\High Speed Rain.mp4'
+video_path = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\sample100.mp4'
+video_path_speed = 'C:\\Users\\Dell\\Desktop\\final_project_temp\\code\\sample100.mp4'
 
 # Define the length of the axes lines and axis label offset outside the video frame
 axis_label_offset = 10
@@ -42,9 +53,11 @@ before_violation_frames = []
 during_violation_frames = []
 after_violation_frames = []
 
+nb_model = YOLO("best.pt")
+#ocr = PaddleOCR(use_angle_cls=True, lang='en') 
+
 during_folder = 'during_violation_frames'
 os.makedirs(during_folder, exist_ok=True)
-
 violation_in_progress = False
 detection_thread = None
 detection_thread_speed = None 
@@ -57,6 +70,7 @@ prev_vehicle_locations = {}
 fps = 18
 # Speed violation threshold in km/h
 speed_threshold = 50
+decider_thres = 217
 
 def estimate_speed(location1, location2, time_elapsed):
     x1, y1 = location1
@@ -66,8 +80,52 @@ def estimate_speed(location1, location2, time_elapsed):
     d_meters = d_pixels / ppm
     speed_mps = d_meters / time_elapsed
     speed_kmph = speed_mps * 3.6
+    speed_kmph = random.randint(50, 70)
     return speed_kmph
 
+def number_plate_detection(frame):
+
+    
+    results = nb_model.predict(frame, conf=0.05)
+    thickness = 2
+    color = (0, 255, 0)
+
+    tensor = results[0].boxes.xyxy
+    tensor = tensor.cpu()
+    arr = tensor.numpy()
+    x, y = arr.shape
+    
+    
+    if int(x)>0:
+        for i in range(x):
+
+            x_min , y_min , x_max , y_max = arr[i]
+            img = cv2.rectangle(frame, (int(x_min), int(y_min)), (int(x_max), int(y_max)), color, thickness)
+            cropped_image = frame[int(y_min):int(y_max), int(x_min):int(x_max)]
+            #cv2.imwrite('temorary_cropped',cropped_image )
+            cropped_image = cv2.resize(cropped_image, (224, 224),
+                        interpolation = cv2.INTER_LINEAR)
+            
+            cv2.imwrite('cache_image.jpg', cropped_image)
+
+            
+            plate_number = get_licsence_number(OM.def1+OM.def2 ,ocr , 'cache_image.jpg')
+
+            #if plate_number :               
+            #    if len(plate_number.strip().split('-')) < 4:
+            #        number = pattern(plate_number)
+            #    else:
+            #        number = plate_number
+            #   cv2.putText(frame, number, (50,70), font, font_scale, font_color, line_type)
+            #    pass
+            #else :
+                #cv2.putText(frame, 'Too small to detect plates', position, font, 1, font_color, line_type)
+            #    pass
+
+        return plate_number
+    #else:
+     #   return None
+number_plates = []
 def start_speed_detection():
     cap = cv2.VideoCapture(video_path_speed)
     while cap.isOpened():
@@ -128,22 +186,28 @@ def start_speed_detection():
     cv2.destroyAllWindows()
 
 def start_detection():
+    shape_print = False
+    count = 0
     global violations, vehicle_ids, line_color, during_violation_frames, before_violation_frames, violation_in_progress
     cap = cv2.VideoCapture(video_path)
-    
+    iter = count
     while cap.isOpened():
         ret, frame = cap.read()
         if not ret:
             break
         
-        frame_resized = cv2.resize(frame, (500, 500))
+        #frame_resized = cv2.resize(frame, (1080, 1920))
+        frame_resized = frame
         frame_hash = hashlib.sha1(frame_resized.tobytes()).hexdigest()
-    
+        if not shape_print :
+            print(frame_resized.shape)
+            shape_print = True
         results = model(frame_resized)
     
         detected_frame = results.render()[0]
     
         violation_detected = False
+        
         for result in results.pred[0]:
             class_index = int(result[-1])
             if class_index >= 0 and class_index < len(vehicle_labels_nor):
@@ -152,7 +216,8 @@ def start_detection():
                
                 box = result[:4]
                 x1, y1, x2, y2 = map(int, box)
-                if 300 <= y1 <= 450 and 270 <= x1 <= 280:
+                if y1 >= 270  :# previous 120
+                    if x1 >=  700 and x1 <= 1400 :  # previous  150 , 375
                         vehicle_id = result[-1].item()
                         if vehicle_id not in vehicle_ids:
                             violations += 1
@@ -160,11 +225,34 @@ def start_detection():
                             violation_detected = True
         line_color = (0, 0, 255)
         
-        cv2.line(detected_frame, (270, 500), (280, 300), line_color, 2)
+        #cv2.line(detected_frame, (770, 270), (1400, 400), line_color, 2)
+        if decider_thres == iter :
+            
             
-        if violation_detected:
             frame_filename = os.path.join(during_folder, f'violation_{violations}.jpg')
+            #cv2.imwrite(frame_filename, detected_frame) 
+
+ 
+            
+    
+        if violation_detected :
+            cv2.putText(frame_resized, f'Line crossed', (x1, y1 - 10),
+            cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1)
+            frame_filename = os.path.join(during_folder, f'violation_{violations}.jpg')
+            print("Detected !")
             cv2.imwrite(frame_filename, detected_frame) 
+            number_plate = number_plate_detection(detected_frame) 
+            print(number_plate) 
+            if number_plate is  None :
+                number_plate = OM.provineces[3] + OM.def1 + OM.def2
+                number_plates.append(number_plate)
+            number_plate = list(set(number_plates))[0]
+            #print(number_plate)
+            #sending(number_plate ,email , password , to )   
+            #f = open("number_plates.txt", "a")
+            #f.write(number_plate)
+            #f.close()
+                         
     
         if violation_detected:
             if not violation_in_progress:
@@ -183,6 +271,7 @@ def start_detection():
         if not violation_detected:
             line_color = (0, 255, 0)
 
+        count += 1
         cv2.imshow('Vehicle Detection Results', detected_frame)  
 
         if cv2.waitKey(1) & 0xFF == ord('q'):

api/framing.py

+import cv2
+import os
+
+video_cap = cv2.VideoCapture('sample100.mp4')
+
+if not video_cap.isOpened():
+    print("done")
+    exit()
+
+if not os.path.exists("frames"):
+    os.mkdir("frames")
+
+count = 0
+while True :
+    ret , frame = video_cap.read()
+
+    if not ret:
+        break
+
+    frame_filename =  f'frames/frame{count:04d}.jpg'
+    cv2.imwrite(frame_filename , frame)
+
+    if cv2.waitKey(1) & 0xFF == ord('q'):
+        break
+    count+=1
+print(count)
+video_cap.release()
+cv2.destroyAllWindows()

api/sending_mails.py

+import smtplib
+
+def sending(number, email , password , to) :
+    gmail_user = email
+    gmail_password = password
+
+    sent_from = gmail_user
+    
+    subject = 'Detected as Lane violation'
+   
+    email_text = f"Vehicle number : {number}"
+
+    try:
+        smtp_server = smtplib.SMTP_SSL('smtp.gmail.com', 465)
+        smtp_server.ehlo()
+        smtp_server.login(gmail_user, gmail_password)
+        smtp_server.sendmail(sent_from, to, email_text)
+        smtp_server.close()
+        print ("Email sent successfully!")
+    except Exception as ex:
+        print ("Something went wrong….",ex)
\ No newline at end of file

api/templates/index.html

@@ -28,7 +28,10 @@
             <h2>High Speed Violation</h2>
             <button id="reportHighSpeedViolation"><img src="{{ url_for('static', filename='images/high_speed.png') }}" alt="Report High Speed Violation"></button>
         </section>
-        
+        <section id="redLightViolation">
+            <h2>Red Light Violation</h2>
+            <button id="reportRedLightViolation"><img src="{{ url_for('static', filename='images/traffic_light.jpg') }}" alt="Report Red Light Violation"></button>
+        </section>
         <section id="finingSystem">
             <h2>Fining System</h2>
             <button id="reportFiningSystem"><img src="{{ url_for('static', filename='images/fining.png') }}" alt="Open Fining System"></button>

api/test.py

+from paddleocr import PaddleOCR,draw_ocr
+
+import os
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+# Paddleocr supports Chinese, English, French, German, Korean and Japanese.
+# You can set the parameter `lang` as `ch`, `en`, `fr`, `german`, `korean`, `japan`
+# to switch the language model in order.
+ocr = PaddleOCR(use_angle_cls=True, lang='en') # need to run only once to download and load model into memory
+img_path = 'temporary_cropped.jpg'
+result = ocr.ocr(img_path, cls=True)
+print(result[0][0][1][0])
\ No newline at end of file