Edit Existing Python Script to convert annotations made in BBLabel -- 2

I have a group of images annotated using BBBox Label / Annotation Tool. [login to view URL] I have multiple classes of object - including "letter", "ornament", "illustration", "portrait", "border", "music", "line"

There is a master convert script available at [login to view URL] which works for 1 class of object (recording everything as class 0). This works perfectly fine for 1 class, and duly converts the bounding box data into the format needed in YOLO, but all classes are represented as 0.

Currently I have 7 classes (and will have more ultimately) and ideally would like a script that could handle that. A script exists that is supposed to support multiple classes. It is contained in a fork in that repo at [login to view URL] However, I can't seem to get this to work.

BBOX Annotation Label .txt file SAMPLE

1 [= 1 ITEM FOUND IN IMAGE]

2 22 564 800 [BOUNDING BOX COORDINATES] illustration [=CLASS OF OBJECT]

I NEED

3 [CLASS OF OBJECT - ILLUSTRATION = NUMBER THREE ON LIST OF OBJECT CLASSES] 0.4921739130434783 0.5055350553505535 0.9773913043478262 0.956949569495695 {COORDINATE DATA REQUIRED BY YOLO}

The second [login to view URL] script listed above is supposed to handle multiple classes but I can't make it work. It may just be a case of explaining to me how to edit the file to get it to function, or it may require extra code to work across multiple classes.

A single .txt annotation file may contain references, for instance, to many letters and an illustration with bounding boxes for each. I have attached a couple of sample files with BBBox .txt annotations.

Image Processing Machine Learning (ML) Python Yazılım Mimarisi

Proje NO: #18467635

Proje hakkında