sqoshi / medusa

Package desired to dataset creation and modification.

Medusa 🐍

Medusa is a python package helpful in dataset creation or modification.

Table Of Contests

• Installation
• Usage
• General
  • Commands

Installation

pip install git+https://github.com/sqoshi/medusa.git

Usage

medusa command --arguments

Example: medusa convert /dataset --img-ext jpg

General

Commands

convert

Convert images to other image extension.

argument	default	description
--input-dir	None	Path to input directory with images.
--output-dir	converted_images	Output directory path (does not have to exist)
--img-ext	png	Target image extension [ jpg,jpeg,webp,png]
first_positional	None	Allow to pass single image.

detect

Find and crop faces in image.

argument	default	description
--input-dir	None	Path to input directory with images.
--output-dir	detected_faces	Output directory path (does not have to exist)
--img-ext	png	Target image extension [ jpg,jpeg,webp,png]
--target-width	160	Width of output image.
--target-height	160	Height of output image.
first_positional	None	Allow to pass single image.

---

During nvidia graphics card setup you may encounter a lot of bugs (TENSORFLOW).

Possible fixes:

1. sudo ln -s /usr/local/cuda-11.0/targets/x86_64-linux/lib/libcusolver.so.11 ~/Documents/medusa/venv/lib/python3.8/site-packages/tensorflow/python/libcusolver.so.11
2. @reboot root for a in /sys/bus/pci/devices/*; do echo 0 | tee -a $a/numa_node; done > /dev/null append to /etc/crontab