app2_correct_segmentation.py

import numpy as np
import json
from skimage import io, color, segmentation, img_as_ubyte, filters, measure
from PIL import Image


import dash_canvas
import dash
from dash.dependencies import Input, Output, State
import dash_html_components as html
import dash_core_components as dcc
import plotly.graph_objs as go

from dash_canvas.utils.parse_json import parse_jsonstring
from dash_canvas.utils.io_utils import image_string_to_PILImage, array_to_data_url
from dash_canvas.utils.image_processing_utils import modify_segmentation


# Image to segment and shape parameters
filename = 'https://upload.wikimedia.org/wikipedia/commons/1/1b/HumanChromosomesChromomycinA3.jpg'
img = io.imread(filename, as_gray=True)
mask = img > 1.2 * filters.threshold_otsu(img)
labels = measure.label(mask)


overlay = segmentation.mark_boundaries(img, labels)
overlay = img_as_ubyte(overlay)

height, width = img.shape[:2]
canvas_width = 500
canvas_height = round(height * canvas_width / width)
scale = canvas_width / width

# ------------------ App definition ---------------------


def title():
    return "Segmentation post-processing"


def description():
    return "Merge or separate labeled regions in order to improve an automatic segmentation"


layout = html.Div([
    html.Div([
    html.H3(children='Manual correction of automatic segmentation'),
    dcc.Markdown('''
        Annotate the picture to delineate boundaries
        between objects (in split mode) or to join objects
        together (in merge mode), then press the 
        "Update segmentation" button to correct
        the segmentation.
    '''),
    html.H5(children='Annotations'),
    dcc.RadioItems(id='mode',
    options=[
        {'label': 'Merge objects', 'value': 'merge'},
        {'label': 'Split objects', 'value': 'split'},
    ],
    value='split',
    # labelStyle={'display': 'inline-block'}
    ),
    html.H5(children='Save segmentation'),
    dcc.RadioItems(id='save-mode',
    options=[
        {'label': 'png', 'value': 'png'},
        #{'label': 'raw', 'value': 'raw'},
    ],
    value='png',
    labelStyle={'display': 'inline-block'}
    ),
    html.A(
        'Download Data',
        id='download-link',
        download="correct_segmentation.png",
        href="",
        target="_blank"
    ),
    dcc.Store(id='cache', data=''),

    ], className="four columns"),
    html.Div([
    dash_canvas.DashCanvas(
        id='canvas_',
        width=canvas_width,
	height=canvas_height,
        scale=scale,
        lineWidth=2,
        lineColor='red',
        image_content=array_to_data_url(overlay),
        goButtonTitle='Update segmentation',
    ),
     ], className="six columns"),
       ])

# ----------------------- Callbacks -----------------------------


def callbacks(app):
    @app.callback(Output('cache', 'data'),
                [Input('canvas_', 'trigger'),],
                [State('canvas_', 'json_data'),
                State('canvas_', 'scale'),
                State('canvas_', 'height'),
                State('canvas_', 'width'),
                State('cache', 'data'),
                State('mode', 'value')])
    def update_segmentation(toggle, string, s, h, w, children, mode):
        print("updating")
        if len(children) == 0:
            labs = labels
        else:
            labs = np.asarray(children)
        with open('data.json', 'w') as fp:
            json.dump(string, fp)
        mask = parse_jsonstring(string, shape=(height, width))
        new_labels = modify_segmentation(labs, mask, img=img, mode=mode)
        return new_labels


    @app.callback(Output('canvas_', 'image_content'),
                [Input('cache', 'data')])
    def update_figure(labs):
        new_labels = np.array(labs)
        overlay = segmentation.mark_boundaries(img, new_labels)
        overlay = img_as_ubyte(overlay)
        return array_to_data_url(overlay)


    @app.callback(Output('download-link', 'download'),
                [Input('save-mode', 'value')])
    def download_name(save_mode):
        if save_mode == 'png':
            return 'correct_segmentation.png'
        else:
            return 'correct_segmentation.raw'


    @app.callback(Output('download-link', 'href'),
                [Input('cache', 'data')],
                [State('save-mode', 'value')])
    def save_segmentation(labs, save_mode):
        new_labels = np.array(labs)
        np.save('labels.npy', new_labels)
        if save_mode == 'png':
            color_labels = color.label2rgb(new_labels)
            uri = array_to_data_url(new_labels, dtype=np.uint8)
            return uri