Skip to content

Latest commit

 

History

History
78 lines (51 loc) · 1.66 KB

README.md

File metadata and controls

78 lines (51 loc) · 1.66 KB

Supervising UI

This project has a web interface to label training data for machine learning task. As of now it can allow you to easily label images with one or many labels

Requires

  1. flask
  2. sqlite3

Install them using pip

Quick Start

Step1 : You need to feed training records

find $HOME -name '*.jpg' > workdir/input.txt

Step 2: Create a settings file that contains labels

Example :

{
  "type": "image-labeling",
  "task": {
	"labels":[
		"class1",
		"class2",
		"class3",
		"class4"
	 ]
  }
}

place this file in a directory with name 'settings.json' Example : workdir/settings.json

Step 3: Start server

Usage

python app.py -h

usage: app.py [-h] [-i INPUT] -w WORK_DIR [-p PORT]

Web UI for Labeling images

optional arguments:
  -h, --help            show this help message and exit
  -i INPUT, --input INPUT
                        Path to to input file which has list of paths, one per
                        line. (Optional)
  -w WORK_DIR, --work-dir WORK_DIR
                        Work Directory. (Required)
  -p PORT, --port PORT  Bind port. (Optional

Example

python app.py -i workdir/input.txt -w workdir -p 8080

Visit http://localhost:8080 on web browser

Screenshots

Image

WARNING:

  • DO NOT MAKE THIS SERVICE ACCESSIBLE FROM PUBLIC INTERNET. Reason: it also acts as a proxy to serve image files to the web browser, it can be tricked by hackers to download other files from file system. So DO NOT OPEN it to public. If you are hosting it within your local network in which the service port is not exposed to internet, then it is fine.