Skip to content

ez-frcnn.annotation

def annotation.encode_image(filepath):

Encodes an image file as a base64 string with a data URI prefix.

Inputs

filepath (str): The path to the image file to be encoded.

Output

str: A base64-encoded string representation of the image, prefixed with "data:image/png;base64," for embedding in HTML or CSS.

Source code in library/annotation.py 
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
def encode_image(filepath):
    """
    Encodes an image file as a base64 string with a data URI prefix.

    Inputs:
        filepath (str): The path to the image file to be encoded.

    Output:
        str: A base64-encoded string representation of the image,
             prefixed with "data:image/png;base64," for embedding in HTML or CSS.

    """
    with open(filepath, 'rb') as f:
        image_bytes = f.read()
    encoded = str(base64.b64encode(image_bytes), 'utf-8')
    return "data:image/png;base64,"+encoded

def annotation.init_annotations(classes):

Initializes an interactive image annotation interface using Jupyter widgets.

Inputs

classes (list of str): A list of class labels that users can assign to bounding boxes.

Output

VBox: A widget container that includes image display, navigation buttons, a label, and a progress bar for annotating images in a directory.

Source code in library/annotation.py 
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
def init_annotations(classes):
    """
    Initializes an interactive image annotation interface using Jupyter widgets.

    Inputs:
        classes (list of str): A list of class labels that users can assign to bounding boxes.

    Output:
        VBox: A widget container that includes image display, navigation buttons,
              a label, and a progress bar for annotating images in a directory.

    """
    global files, annotations, current_index  # Ensure these are accessible globally
    path = './annotations/'
    current_index = 0  # Initialize the index variable
    annotations = {}
    image_extensions = ['jpg', 'jpeg', 'gif', 'bmp', 'tiff', 'webp']
    all_extensions = image_extensions + [ext.upper() for ext in image_extensions]  # Add uppercase versions
    files = glob.glob(f"{path}/*.png")
    for extension in all_extensions:
        files.extend(glob.glob(f"{path}/*.{extension}"))
    files = [os.path.basename(image_path) for image_path in files]  # More portable than split('/')
    if not files:
        print("No images found in the specified directory.")
        return
    im = encode_image(os.path.join(path, files[0]))
    w_bbox = BBoxWidget(
        image=im,
        classes=classes,
        hide_buttons=True
    )

    # Initialize the label widget here to update it in the functions
    label = widgets.Label(value=files[0])

    # Define the function to update the image and label
    def update_image_and_annotations():
        nonlocal label  # Indicates that we're using the `label` defined outside this function
        global current_index
        if current_index < len(files):
            image_file = encode_image(os.path.join(path, files[current_index]))
            w_bbox.image = image_file  # Update the image in the widget
            w_progress.value = current_index  # Update the progress bar
            w_progress.description = f"{current_index + 1} / {len(files)}"  # Update the progress bar name with index/total
            label.value = files[current_index]  # Update the label
        else:
            print("Reached the end of the file list.")

    def go_back(_):
        global current_index
        if current_index > 0:
            current_index -= 1
            w_bbox.bboxes = []
            update_image_and_annotations()

    def skip(_):
        global current_index
        if current_index < len(files) - 1:
            current_index += 1
            w_bbox.bboxes = []
            update_image_and_annotations()


    def submit(_):
        global annotations, current_index
        image_file = files[w_progress.value]
        path = './annotations/'
        #print(image_file)

        image = cv2.imread(os.path.join(path, image_file))
        height, width, _ = image.shape
        # save annotations for current image
        annotations[image_file] = w_bbox.bboxes
        if annotations[image_file]:
            annotations_for_image = {'annotation': {'object': annotations[image_file]}}
        else:
            # No annotations, create a 'background' label with full image bounding box
            background_annotation = {

                    'x': 0,
                    'y': 0,
                    'width': width,
                    'height': height,
                    'label':'background'

            }
            annotations_for_image = {'annotation': {'object': background_annotation}}

        xml_string = xmltodict.unparse(annotations_for_image, pretty=True)
        with open(path + image_file[:-4] + '.xml', 'w') as f:
            #json.dump(annotations[image_file], f, indent=4)
            f.write(xml_string)
        # move on to the next file
        skip(_)

    def on_reset_clicked(b):
        w_bbox.bboxes = []  # Clears the drawn boxes in your widget

    # Connect buttons to their callbacks
    go_back_button = widgets.Button(description="Go Back", layout=Layout(width='auto'))
    go_back_button.on_click(go_back)

    skip_button = widgets.Button(description="Skip", layout=Layout(width='auto'))
    skip_button.on_click(skip)

    reset_button = widgets.Button(description="Reset", layout=Layout(width='auto'))
    reset_button.on_click(on_reset_clicked)

    submit_button = widgets.Button(description="Submit", layout=Layout(width='auto'))
    submit_button.on_click(submit)

    # Progress bar to show how far we got
    # Progress bar to show how far we got, moved slightly to the left using margin
    w_progress = widgets.IntProgress(
        value=0, 
        max=len(files)-1, 
        description=f'1 / {len(files)}',
        layout=widgets.Layout(width='287px', margin='0 0 0 -40px')  # Adjust width and left margin
    )


    # Combine widgets into containers
    navigation_buttons = widgets.HBox([go_back_button, skip_button, submit_button, reset_button], layout=Layout(margin='0 0 0 3px'))
    w_progressLabeled = widgets.HBox([w_progress, label])

    w_container = widgets.VBox([
        w_progressLabeled,
        navigation_buttons,
        w_bbox,

    ])

    return w_container

def annotation.split_images_and_xml(source_folder, train_folder='./images/train/', test_folder='./images/test/', test_ratio=0.1):

Randomly splits a set of images and corresponding XML annotation files into training and testing folders. Generates individual CSV annotation files for each image based on its XML.

Inputs

source_folder (str): Folder containing the original image and XML files. train_folder (str): Destination folder for training images and annotations. Default is './images/train/'. test_folder (str): Destination folder for testing images and annotations. Default is './images/test/'. test_ratio (float): Proportion of the dataset to allocate to the test set. Default is 0.1 (10%).

Output

None

Source code in library/annotation.py 
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
def split_images_and_xml(source_folder, train_folder='./images/train/', test_folder='./images/test/', test_ratio=0.1):
    """
    Randomly splits a set of images and corresponding XML annotation files into training and testing folders.
    Generates individual CSV annotation files for each image based on its XML.

    Inputs:
        source_folder (str): Folder containing the original image and XML files.
        train_folder (str):  Destination folder for training images and annotations. Default is './images/train/'.
        test_folder (str):   Destination folder for testing images and annotations. Default is './images/test/'.
        test_ratio (float):  Proportion of the dataset to allocate to the test set. Default is 0.1 (10%).

    Output:
        None
    """
    # Ensure the train and test folders exist
    os.makedirs(train_folder, exist_ok=True)
    os.makedirs(test_folder, exist_ok=True)

    image_extensions = ['jpg', 'jpeg', 'gif', 'bmp', 'tiff', 'webp']
    all_extensions = image_extensions + [ext.upper() for ext in image_extensions]  # Add uppercase versions
    # Find all image files in the source folder. .
    image_files = glob.glob(os.path.join(source_folder, "*.png"))
    path = './annotations'
    for extension in all_extensions:
        image_files.extend(glob.glob(f"{path}/*.{extension}"))

    # Shuffle the image files to ensure random selection
    random.shuffle(image_files)

    # Calculate the number of files to allocate to the test set
    num_test_files = int(len(image_files) * test_ratio)

    # Split the files into training and testing sets
    test_files = image_files[:num_test_files]
    train_files = image_files[num_test_files:]

    # Function to convert XML annotations to CSV format
    def xml_to_csv(xml_file, output_csv_file):
        with open(xml_file) as xml_fd:
            annotation_data = xmltodict.parse(xml_fd.read())
            objects = annotation_data.get('annotation', {}).get('object', [])

            # Handle case if only one object exists (not a list)
            if not isinstance(objects, list):
                objects = [objects]

            csv_rows = []
            for obj in objects:
                if obj is not None:
                    class_name = obj['label']
                    x = obj['x']
                    y = obj['y']
                    width = obj['width']
                    height = obj['height']
                    csv_rows.append([class_name, x, y, width, height])

            # Convert rows to a pandas DataFrame and save to CSV
            df = pd.DataFrame(csv_rows, columns=['class', 'xmin', 'ymin', 'xmax', 'ymax'])
            df.to_csv(output_csv_file, index=False)

    # Copy the files to their respective folders and generate CSV annotations
    for file_path in test_files:
        shutil.copy(file_path, test_folder)
        xml_path = os.path.splitext(file_path)[0] + '.xml'
        if os.path.exists(xml_path):
            shutil.copy(xml_path, test_folder)
            csv_path = os.path.splitext(file_path)[0] + '.csv'
            xml_to_csv(xml_path, os.path.join(test_folder, os.path.basename(csv_path)))

    for file_path in train_files:
        shutil.copy(file_path, train_folder)
        xml_path = os.path.splitext(file_path)[0] + '.xml'
        if os.path.exists(xml_path):
            shutil.copy(xml_path, train_folder)
            csv_path = os.path.splitext(file_path)[0] + '.csv'
            xml_to_csv(xml_path, os.path.join(source_folder, os.path.basename(csv_path)))