CompareAnnotations is a simple widget that allows you to compare different annotations for one image. It can be useful for comparing annotations of different applied NN models. Widget doesn't support Keypoints shape. This widget is based on GridGallery widget.
If no image is given, this message will be displayed.
type:str
default value:"No image was provided"
compare_annotations = CompareAnnotations(columns_number=3, empty_message="Set image URL and annotations")
widget_id
ID of the widget.
type:str
default value:None
Methods and attributes
Attributes and Methods
Description
image_url()
Returns current image URL
set_image_url()
Set current image by URL
append()
Add annotations to gallery
is_empty()
Checks whether image is set or not
clean_up()
Clean up widget from image and annotations
Mini App Example
In this example, we will use CompareAnnotations widget to display different annotations for one image. We will modify and duplicate an existing image annotation and slightly alter the labels on it, to make them different. All labels will be converted to shape Rectangle.
You can find this example in our Github repository:
import os
from random import choice
import supervisely as sly
from dotenv import load_dotenv
from supervisely.app.widgets import (
Button,
Card,
Container,
CompareAnnotations,
)
from supervisely import Annotation, Rectangle
Init API client
First, we load environment variables with credentials and init API for communicating with Supervisely Instance:
if sly.is_development():
load_dotenv("local.env")
load_dotenv(os.path.expanduser("~/supervisely.env"))
api = sly.Api()
Get project ID and dataset ID
This app requires that you project have at least 2 images in it. If you don't have any projects, you can get one from Supervisely Ecosystem.
Option 1. Collect image names, URLs, annotations and generate annotation names from server
image_names = []
image_urls = []
image_anns = []
for idx in range(len(images_infos)):
image_names.append(images_infos[idx].name)
image_urls.append(images_infos[idx].full_storage_url)
image_anns.append(
sly.Annotation.from_json(data=anns_infos[idx].annotation, project_meta=project_meta)
)
ann_names = [f"Model inference {idx+1}" for idx in range(len(image_anns))]
Option 2. You can also serve images from your local machine, using a static directory
Sort lists of images and annotations to make sure they are in the correct order.
static_dir = Path("compare data/005_compare_annotations/images")
ann_dir = Path("compare data/005_compare_annotations/annotations")
image_urls = sorted([
f"/static/{get_file_name_with_ext(path)}"
for path in static_dir.iterdir()
if path.is_file()
])
local_annotations = sorted([str(path) for path in ann_dir.iterdir() if path.is_file()])
Read annotations from local .json files
image_anns = []
for local_ann in local_annotations:
ann = sly.Annotation.from_json(
data=sly.json.load_json_file(local_ann), project_meta=project_meta
)
image_anns.append(ann)
Create a function to modify the annotation
def ann_to_bbox(annotation: Annotation):
labels = []
mapping = {}
for label in annotation.labels:
if label.obj_class.name not in mapping:
new_obj_class = sly.ObjClass(label.obj_class.name, Rectangle)
mapping[label.obj_class] = new_obj_class
label = label.scale(choice(range(90, 100)) / 100)
labels.append(label)
annotation = annotation.clone(labels=labels)
annotation = annotation.to_detection_task(mapping)
return annotation
Iterate over column numbers and append modified annotations to widget
We iterate over column numbers for demo purposes to show all annotations in one line.
for i in range(compare_annotations.columns_number):
ann = ann_to_bbox(image_anns[0])
compare_annotations.append(
annotation=ann,
title=ann_names[i],
column_index=i,
)
Create buttons for changing current image and cleaning up widget
