Commit 80202410 authored by narugo1992's avatar narugo1992
Browse files

dev(narugo): update README.md

parent 0096203d

README.md

+33 −0
Original line number Diff line number Diff line
@@ -41,6 +41,13 @@ to [Installation](https://deepghs.github.io/imgutils/main/tutorials/installation 


## Supported or Developing Features



* [Tachie(差分) Detection and Clustering](https://github.com/deepghs/imgutils#tachie%E5%B7%AE%E5%88%86-detection-and-clustering)

* [Object Detection](https://github.com/deepghs/imgutils#object-detection)

* [Monochrome Image Detection](https://github.com/deepghs/imgutils#monochrome-image-detection)

* [Truncated Image Check](https://github.com/deepghs/imgutils#truncated-image-check)

* [Image Tagging](https://github.com/deepghs/imgutils#image-tagging)

* [Character Extraction](https://github.com/deepghs/imgutils#character-extraction)



### Tachie(差分) Detection and Clustering



For the dataset, we need to filter the differences between the tachie(差分). As shown in the following picture
@@ -59,6 +66,32 @@ print(lpips_clustering(images)) # -1 means noises, the same as that in sklearn 
# [0, 0, 0, 1, 1, -1, -1, -1, -1]

```



### Object Detection



Currently, object detection is supported for anime faces and person, as shown below



* Face Detection



![face detection](https://deepghs.github.io/imgutils/main/_images/face_detect.dat.svg)



* Person Detection



![person detection](https://deepghs.github.io/imgutils/main/_images/person_detect.dat.svg)



Based on practical tests, face detection currently has a very stable performance and can be used for automation tasks.

However, person detection is still being further iterated and will focus on enhancing detection capabilities for

artistic illustrations in the future.



### Edge Detection / Lineart Generation



Anime images can be converted to line drawings using the model provided

by [patrickvonplaten/controlnet_aux](https://github.com/patrickvonplaten/controlnet_aux), as shown below.



![edge example](https://deepghs.github.io/imgutils/main/_images/edge.dat.svg)



It is worth noting that the `lineart` model may consume more computational resources, while `canny` is the fastest but

has average effect. Therefore, `lineart_anime` may be the most balanced choice in most cases.



### Monochrome Image Detection



When filtering the crawled images, we need to remove monochrome images. However, monochrome images are often not simply

imgutils/config/meta.py

+2 −2
Original line number Diff line number Diff line
@@ -7,10 +7,10 @@ Overview: 
__TITLE__ = 'imgutils'



#: Version of this project.

__VERSION__ = '0.0.1'

__VERSION__ = '0.0.2'



#: Short description of the project, will be included in ``setup.py``.

__DESCRIPTION__ = 'Utilities of images.'

__DESCRIPTION__ = 'A convenient and user-friendly anime-style image data processing library that integrates various advanced anime-style image processing models.'



#: Author of this project.

__AUTHOR__ = 'narugo1992'