stef_buwalda/ProjectIOT
1
0
Fork 0
mirror of https://github.com/StefBuwalda/ProjectIOT.git synced 2025-11-01 20:29:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

Only OCR the biggest bounding box

Browse Source
This commit is contained in:
Stef
2025-06-12 11:34:42 +02:00
parent 15d9d33ba8
commit d422268a03
1 changed files with 32 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
ANPR.py
View File

@@ -1,4 +1,5 @@
from ultralytics import YOLO from ultralytics import YOLO
from ultralytics.engine.results import Boxes
from PIL import Image from PIL import Image
import numpy as np import numpy as np
import easyocr import easyocr
@@ -11,31 +12,35 @@ img = Image.open("test.jpg")
results = car_model.predict(source=img) results = car_model.predict(source=img)
for r in results: cars: list[tuple[int, tuple[int, int, int, int]]] = []
for box in r.boxes:
cls_id = int(box.cls[0])
cls_name = r.names[cls_id]
if cls_name == "car":
conf = float(box.conf[0])
x1, y1, x2, y2 = map(int, box.xyxy[0])
print(f"{cls_name} ({conf:.2f}) at [{x1},{y1},{x2},{y2}]")
# Crop Image # Filter out the cars and calculate box size
cropped_img = img.crop((x1, y1, x2, y2)) for r in results:
if r.boxes:
for box in r.boxes:
cls_name = r.names[int(box.cls[0])]
if cls_name == "car":
x1, y1, x2, y2 = map(int, box.xyxy[0])
size = (x2 - x1) ** 2 + (y2 - y1) ** 2
cars.append((size, (x1, y1, x2, y2)))
# Get the biggest car box
size, corners = max(cars, key=lambda x: x[0])
# Crop biggest car
cropped_img = img.crop(corners)
cropped_img.save("car_crop_pillow.jpg") cropped_img.save("car_crop_pillow.jpg")
results2 = plate_model.predict(source=cropped_img)
for x in results2: # Search for license plates in car box and OCR all
for box2 in x.boxes: results = plate_model.predict(source=cropped_img)
cls_id = int(box2.cls[0]) for r in results:
cls_name = x.names[cls_id] if r.boxes:
for box in r.boxes:
cls_name = r.names[int(box.cls[0])]
if cls_name == "License_Plate": if cls_name == "License_Plate":
conf = float(box2.conf[0]) x1, y1, x2, y2 = map(int, box.xyxy[0])
x1, y1, x2, y2 = map(int, box2.xyxy[0])
print(
f"{cls_name} ({conf:.2f}) at [{x1},{y1},{x2},{y2}]"
)
lp_img = cropped_img.crop((x1, y1, x2, y2)) lp_img = cropped_img.crop((x1, y1, x2, y2))
lp_img.save("license_plate.jpg") lp_img.save("license_plate.jpg")
lp_np = np.array(lp_img) lp_np = np.array(object=lp_img)
result3 = reader.readtext(lp_np) result = reader.readtext(image=lp_np)
print(result3) print(result)