get rid of halftone grid

[github-actions]

backsteinexpressionismus/Source Files/Mocks/generator/lib/ultralyticsextractor.py

Line 12 in 2a21f01

# TODO: get rid of halftone grid

import os
from PIL import Image
from ultralytics import YOLO
from .ultralyticslayout import UltralyticsLayout
from .ultralyticsoiv7classifier import SCORES
from .extractor import Extractor
from .layout import Block
from termcolor import cprint

# TODO: Annotate
# TODO: Resize
# TODO: get rid of halftone grid

lib_dir = os.path.dirname(os.path.realpath(__file__))
# TODO: Add Model Management
model_path = "../models/yolov8x-oiv7.pt"
model_path = os.path.join(lib_dir, model_path)


class UltralyticsExtractor(Extractor, UltralyticsLayout):
    model = YOLO(model_path)
    #print(repr(model))

    def __init__(self, image, max_px=640, additional_classes=[]):
        self.orig_image = image
        self.additional_classes = additional_classes
        
        #self.image = image.thumbnail((max_px,max_px), Image.Resampling.LANCZOS)
        #self.image = UltralyticsExtractor.preprocess_scale(image, max_px)
        self.image = image
        #self.image.show()
        self.detected_blocks = []
        #super().__init__(image, name, border)

    def preprocess_scale(image, max_px):
        width, height = image.size
        if width > max_px or height > max_px:
            image = image.copy()
            image.thumbnail((max_px,max_px), Image.Resampling.LANCZOS)
        return image

    def annotate(self, logger=None):
        self.detect(model=UltralyticsExtractor.model)
        self.detected_blocks = []
        for result in self.results:
            names = result.names
            for box in result.boxes:
                t = names[int(box.cls)]
                if t in SCORES['EXTERIOR']["classes"]:
                    coords = box.xyxy.numpy().flatten().astype(int)
                    b = Block(coords[0], coords[1], coords[2], coords[3], t)
                    self.detected_blocks.append(b)
                    if logger is not None:
                        logger.info(f"Extracting detected object {t} at {coords}")
                    cprint(f"{t}: {coords}", "white")
        map(lambda x: cprint(x.__dict__, 'white'), self.detected_blocks)
        #for result in self.results:
        #    print(result)

    def scale(self):
        raise NotImplementedError()