image dataset eda tool to check basic infos of images.
Project description
basic-image-eda
A simple eda tool to check basic infos of images under a directory(images are found recursively). This tool was made to quickly check info and prevent mistakes on reading, resizing, and normalizing images as inputs for neural networks. It can be used when first joining an image competition or training CNNs with images!
Notes:
- All images are converted to 3-channel(rgb) images. When images that have various channels are mixed, results can be misleading.
- uint8 and uint16 data types are supported. If different data types are mixed, error occurs.
Installation
pip install basic-image-eda
prerequisites:
- opencv-python
- numpy
- matplotlib
- tqdm
Usage(CLI/Code)
CLI
simple one line command!
basic-image-eda <data_dir>
or
basic-image-eda <data_dir> --extensions png jpg --threads 12 --dimension_plot False --channel_hist True --nonzero
Options:
-e --extensions target image extensions.(default=['png', 'jpg', 'jpeg'])
-t --threads number of multiprocessing threads. if zero, automatically counted.(default=0)
-d --dimension_plot show dimension(height/width) scatter plot.(default=True)
-c --channel_hist show channelwise pixel value histogram. takes much longer time.(default=False)
-n --nonzero calculate values only from non-zero pixels of the images.(default=False)
-V --version show version.
Code
from basic_image_eda import BasicImageEDA
if __name__ == "__main__": # for multiprocessing
data_dir = "./data"
# below are default values.
extensions = ['png', 'jpg', 'jpeg']
threads = 0
dimension_plot = True
channel_hist = False
nonzero = False
BasicImageEDA.explore(data_dir, extensions, threads, dimension_plot, channel_hist, nonzero)
Results
Results on celeba dataset (test set)
found 19962 images.
Using 12 threads.
*--------------------------------------------------------------------------------------*
number of images | 19962
dtype | uint8
channels | [3]
extensions | ['jpg']
min height | 85
mean height | 591.8215108706543
max height | 5616
min width | 85
mean width | 490.2976655645727
max width | 5616
mean height/width ratio | 1.207065732587525
recommended input size | [592 488] (h x w, multiples of 8)
recommended input size | [592 496] (h x w, multiples of 16)
recommended input size | [576 480] (h x w, multiples of 32)
channel mean(0~1) | [0.49546506 0.42573904 0.39331011]
channel std(0~1) | [0.32161251 0.30237885 0.30192492]
*--------------------------------------------------------------------------------------*
Results on NIH Chest X-ray dataset (images_001.tar.gz)
found 4999 images.
Using 12 threads.
*--------------------------------------------------------------------------------------*
number of images | 4999
dtype | uint8
channels | [1, 4]
extensions | ['png']
min height | 1024
mean height | 1024.0
max height | 1024
min width | 1024
mean width | 1024.0
max width | 1024
mean height/width ratio | 1.0
recommended input size | [1024 1024] (h x w, multiples of 8)
recommended input size | [1024 1024] (h x w, multiples of 16)
recommended input size | [1024 1024] (h x w, multiples of 32)
channel mean(0~1) | [0.51725466 0.51725466 0.51725466]
channel std(0~1) | [0.25274113 0.25274113 0.25274113]
*--------------------------------------------------------------------------------------*
License
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
Hashes for basic_image_eda-0.0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 0910c9090f80d2796cd0d3ea817b2268a796fabdaf2e132cd95e465faa5b0137 |
|
MD5 | a07ba747a2168e6b184e0e006f3c83fd |
|
BLAKE2b-256 | 294ff442861da1a9258173395b4811f0dcbde13f719be1f43888794e95c993e9 |