chacoff/vott-csv-to-darknet-format

If you are using VoTT to annotate RGB images and later on you want to feed a Yolo, you might need to convert those to Darknet format. In my case, i figured was easier to export VoTT annotations as CSV and later on make a small script to get theses in Darknet format

VoTT-CSV-to-Darknet-Format

If you are using VoTT to annotate RGB images and later on you want to feed a Yolo, you might need to convert those to Darknet format. In my case, I figured was easier to export VoTT annotations as CSV and later make a small script to get these in Darknet format

Description

• you will need the folder where the images are because we need to open each one of them in order know its size before we can calculate xcen, ycen, height and width for darknet format
• you can choose where to store the *.txt files, either in a brand new folder or in the same folder than the images

Constrainsts

The way i am rounding the floating numbers and storing the txt files might not be the most efficients. With an Alienware i7-8750H, 16GB of RAM and RTX2070Q, it takes about 2min to go through 1141 images, most of them in 1920x1080px.

If you have any suggestion, i'll be very happy to hear them.