yolo2voc2coco

labels: txt-->xml-->json

This project support to Nanodet project to make official labels.

1.Nanodet project address

https://github.com/RangiLyu/nanodet

2.making labels means

2.1 alfred tool address

https://github.com/jinfagang/alfred

2.2 download and install

git clone https://github.com/jinfagang/alfred

sudo pip3 install alfred-py #if using python3

sudo pip install alfred-py #otherwise

2.2.3 yolo-->voc-->coco

2.2.3.1 label: txt-->xml

alfred data yolo2voc --image_dir /Path to yours/coco/images/val2017 --text_dir /Path to yours/coco/labels/val2017 --class_file /Path to yours/classes.txt

or

python3 alfred/modules/data/yolo2voc /Path to your image_dir /Path to your text_dir /Path to your class_file

See the classes.txt file for details, particular attention as below:

such as 'traffic light' category, not same as 'traffic_light' category in nanodet/config/nanodet-m.yml

2.2.3.2 label: xml-->json

python3 alfred/modules/data/voc2coco.py /Path to yours/valxml /Path to yours/val2017.json

or

alfred data voc2coco --xml_dir /Path to yours/valxml/ --index_1 0 #default annotations_coco.json

valxml: the folder to store xml format of voc label file

val2017.json: the json format of coco label file

voc2coco.py: add pre-define category, be sure the labels match.

Multi class label change to one category label

modified voc2coco.py, made other category object bbox not write,as following:

in voc2coco.py

new_id = len(categories)+1 if index_1 else len(categories)

categories[category] = new_id

Remove above 2 lines and replace them with continue.

Negative sample

image and label 1.both names needs to be the same; 2.the label file content is empty.

If I can help you, please give me a star 🌟 🌟 🌟

my csdn: https://blog.csdn.net/yyqq7226741/article/details/110426728