Deep learning Python image marker tool Labeltool__python

Depth learning training needs to mark the location and category of images, used before is called Bbox-label-tool-master, the encounter of large images is not complete, there is no adaptive scaling, this is the improved Python script.

Directory structure: Picture directory name images, label directory name labels, image directory under various categories of directory name to 001,002,003,... The format name.

This is the screenshot when the Labeltool (Python main.py) is running, and the image frame records the proportional value,

Shortcut key a,d before and after the image, ESC cancel the current box

Python script main.py:

[Python]  View Plain  copy   # -*- coding:utf-8 -*-   #----------------------------- --------------------------------------------------   # name:         Object bounding box label tool   # purpose:      Label object bboxes for ImageNet Detection data   #  author:      qiushi   # created:     06/ 06/2014      #   #----------------------------------------------------------------- --------------   from __future__ import division   from tkinter  import *   import tkmessagebox   from pil import image,  imagetk   import os   import glob   import random    & nbsp; w0 = 1;  #图片原始宽度    h0 = 1;  #图片原始高度       # colors  for the bboxes   colors = [' red ',  ' blue ',  ' yellow ',  ' pink ',   ' cyan ',  ' green ',  ' black ']   # image sizes for the examples    size = 256, 256      #指定缩放后的图像大小    dest_size =  500, 500       class labeltool ():        def __init__ (self, master):           # set  up the main frame           self.parent  = master           self.parent.title ("Labeltool")            self.frame = frame (self.parent)             self.frame.pack (fill=both, expand=1)             self.parent.resizable (width = true, height = true )               # initialize global  state