import os
# 圖像文件夾路徑
image_folder = 'D:'
# 標(biāo)注文件即txt文件路徑
txt_folder = 'D:'
# 基于上述兩個(gè)路徑檢索圖像和txt即yolo格式的標(biāo)注文件
image_files = [f.split('.')[0] for f in os.listdir(image_folder) if f.lower().endswith(('jpg', 'jpeg', 'png', 'bmp'))]
txt_files = [f.split('.')[0] for f in os.listdir(txt_folder) if f.lower().endswith('txt')]
# 遍歷圖像文件名
for img_name in image_files:
    # 如果某圖像沒(méi)有對(duì)應(yīng)的標(biāo)簽文件即txt文件
    if img_name not in txt_files:
        # 寫(xiě)入txt文件
        txt_content = """0 0.808621 0.945652 0.072414 0.073913
                        0 0.250000 0.604348 0.093103 0.078261
                        0 0.660345 0.584783 0.086207 0.100000
                        0 0.613793 0.160870 0.117241 0.139130
                        0 0.293103 0.184783 0.103448 0.117391"""
        with open(os.path.join(txt_folder, img_name + '.txt'), 'w') as f:
            f.write(txt_content)
print("Process complete.")

import os
from ultralytics import YOLOv10
# Folder paths
image_folder = 'D:'
txt_folder = 'D:'
# Load a pretrained YOLOv10n model
model = YOLOv10("\weights\\best.pt")
image_files = [f for f in os.listdir(image_folder) if f.lower().endswith(('jpg', 'jpeg', 'png', 'bmp'))]
# Perform object detection on each image file
for img_name in image_files:
    img_path = os.path.join(image_folder, img_name)
    # Perform prediction
    results = model.predict(img_path)
    # Extract detection results (assuming results[0] contains the detection)
    detections = results[0].boxes.xywh  # Get bounding box coordinates (xywh format)
    # If no .txt file exists, create a new one
    txt_path = os.path.join(txt_folder, img_name.split('.')[0] + '.txt')
    if not os.path.exists(txt_path):  # Check if .txt file already exists
        # Write detection results to the .txt file
        with open(txt_path, 'w') as f:
            for detection in detections:
                # Assuming you want the format: class_id x_center y_center width height (normalized)
                # Convert coordinates from pixels to normalized values by dividing by image width/height
                x_center, y_center, width, height = detection
                x_center /= results[0].orig_img.shape[1]  # Normalize by image width
                y_center /= results[0].orig_img.shape[0]  # Normalize by image height
                width /= results[0].orig_img.shape[1]  # Normalize by image width
                height /= results[0].orig_img.shape[0]  # Normalize by image height
                # Write to file (Assuming class id is 0 here, change based on your model)
                f.write(f"0 {x_center} {y_center} {width} {height}\n")
print("Process complete.")