chenghanc / crop-images

Cropping images using trained darknet model

Crop Images

Cropping images using trained darknet model

Usage:

• Extract detected objects information

  • class names, confidence and bbox coordinates
  • Extract the information by running the command

./darknet detector test custom.data custom.cfg custom.weights -thresh 0.4 -dont_show -ext_output < val.txt > result-dr.txt

• Copy the file

  • Copy the file result-dr.txt to the working directory

• Copy the image files

  • Copy the image files to the folder val-images/

• Remove lines from the text file containing specific words

  • Remove Enter by typing grep -vE "(Enter)" result-dr.txt > result4.txt
  • Remove Detection by typing grep -vE "(Detection)" result4.txt > result.txt

• Run the code

  • python parsing.py