YOLOV8：如何保存模型的输出

Question

这是代码

from ultralytics import YOLO
license_plate_detector = YOLO('./model/best.pt')
license_plates = license_plate_detector('./42.jpg')

这是输出

640x608 1 number-plate, 342.0ms
Speed: 12.4ms preprocess, 342.0ms inference, 3.0ms postprocess per image at shape (1, 3, 640, 608)

我想将此输出转换为图像并将其保存以与 esyocr 一起使用

该类没有任何保存方法，那么如何执行此操作

Answer 1

我想，这会对你有帮助。

from PIL import Image
from ultralytics import YOLO
import easyocr

license_plate_detector = YOLO('./model/best.pt')

input_image = Image.open('./42.jpg')

detections = license_plate_detector(input_image)

license_plate_boxes = detections.xyxy[0].cpu().numpy()

reader = easyocr.Reader(['en'])

for i, box in enumerate(license_plate_boxes):
    x1, y1, x2, y2, conf, cls = box
    license_plate = input_image.crop((x1, y1, x2, y2))
    
    plate_filename = f'license_plate_{i}.jpg'
    license_plate.save(plate_filename)

    results = reader.readtext(plate_filename)
    print(f"License Plate {i+1} Text: {results[0][1]}")