A way to label multi label image datasets in jupyter
Project description
🐦 multi_class_pigeon - Quickly annotate data on Jupyter
This repo is a simple multiclass image extenstion of the pigeon repo by @agermanidis located here. Credit goes to them for building the annotate function for images, text, and regression for multi-class type problems. All I am doing is extending it to multi-label and multi-task type problems.
Notes
As of now I just have this as a repo... so to use it you can clone it and either run notebooks from inside it and use the below format or modify the imports to point at the directory containing the annotation script.
Examples
from pigeon import multi_label_annotate
from IPython.display import display, Image
annotations = multi_label_annotate(
['assets/altera.jpg', 'assets/chibi_gil.jpg','assets/chibi_saber.jpg'],
options={'cute':['yes','no'], 'saber':['yes','no'],'colors':['blue','gold','white','red']},
display_fn=lambda filename: display(Image(filename))
)
Preview:
Additional Notes
I added some additional buttons as well to the multi-label script. Since we want to be able to select more than one category I added a few extra buttons.
-
done
button to press after all relevant fields have been clicked. -
back
button in case you want to go back to a previous image. WARNING this will erase that item from the dictionary so you will have to click through all the classes you want to mark for that previous image again. -
clear current
will delete the current image from the dictionary on the back end so you can re enter the classes in case you messed one up.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distributions
Built Distribution
File details
Details for the file multi_label_pigeon_jupyter-0.1.0-py3-none-any.whl
.
File metadata
- Download URL: multi_label_pigeon_jupyter-0.1.0-py3-none-any.whl
- Upload date:
- Size: 4.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/1.13.0 pkginfo/1.4.2 requests/2.19.1 setuptools/40.2.0 requests-toolbelt/0.9.1 tqdm/4.26.0 CPython/3.7.2
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | f09a745c68f7396e8ab39313d16f718dc977cbecad5a707844a109043c84c4f6 |
|
MD5 | 5e04ca44a5d023ac0a071c5233e86c4b |
|
BLAKE2b-256 | c6afb750d0076a9d389c38015171f4ef76403037de024ca79ffb5acbc753e0a8 |