<aside> <img src="/icons/warning_gray.svg" alt="/icons/warning_gray.svg" width="40px" /> The following guide is created on macOS so you might need to adapt a few of them on your platform. Feel free to ask me questions via email.

</aside>

<aside> <img src="/icons/light-bulb_gray.svg" alt="/icons/light-bulb_gray.svg" width="40px" /> Follow the previous guide (Open and edit an annotated Labelme JSON file) to annotate file or download it from below:

<aside> <img src="/icons/document_gray.svg" alt="/icons/document_gray.svg" width="40px" /> The edited file:

dogs_edited.json

</aside>

</aside>

1. Make sure you have dogs_edited.json file.

   $ ls
   dogs_edited.json
   

2. Install required software following

   Install labelme Python package

3. Create a Python file load_annotated_file.py:

   import base64
   import json
   
   import labelme
   import matplotlib.pyplot as plt
   import numpy as np
   
   # load annoated file
   json_file = "dogs_edited.json"
   with open(json_file, "r") as f:
       data = json.load(f)
   
   # load image from data["imageData"]
   image = labelme.utils.img_b64_to_arr(data["imageData"])
   print("image:", image.shape, image.dtype)
   
   # load label_names, label, label_points_xy from data["shapes"]
   unique_label_names = ["_background_"] + sorted(
       set([shape["label"] for shape in data["shapes"]])
   )
   label = np.zeros(image.shape[:2], dtype=np.int32)
   label_names = []
   label_points_xy = []
   for shape in data["shapes"]:
       label_id = unique_label_names.index(shape["label"])
   
       label_names.append(shape["label"])
   
       mask = labelme.utils.shape_to_mask(
           img_shape=image.shape[:2],
           points=shape["points"],
           shape_type=shape["shape_type"],
       )
       label[mask] = label_id
       label_points_xy.append(shape["points"])
   print("label:", label.shape, label.dtype)
   print("label_names:", label_names)
   
   # visualize
   colors = "rgbymc"
   #
   plt.subplot(1, 2, 1)
   plt.imshow(image)
   plt.imshow(label, alpha=0.5)
   #
   plt.subplot(1, 2, 2)
   plt.imshow(image)
   for i, (label_name, label_points_xy_i) in enumerate(zip(label_names, label_points_xy)):
       label_id = unique_label_names.index(label_name)
       label_points_xy_i = np.array(label_points_xy_i)
       plt.plot(
           label_points_xy_i[:, 0],
           label_points_xy_i[:, 1],
           marker="o",
           color=colors[label_id % len(colors)],
           label=label_name if label_name not in label_names[:i] else None,
       )
   plt.legend()
   #
   plt.tight_layout()
   plt.suptitle(json_file)
   plt.show()
   

   

   

   <aside> <img src="/icons/document_gray.svg" alt="/icons/document_gray.svg" width="40px" /> The Python file:

   load_annotated_file.py

   </aside>

Next Guide:

Export a Labelme JSON file to PNG and etc.