Description: Repository to explore any annotated dataset and create COCO styled JSON files to train object detectors on the annotated dataset. The repository supports XML and TXT files.
Images: 4080
Categories: 20
Total instances: 32964
python main.py --root_dir /data/annotated_dataset --output_dir /data/annotated_dataset/annotations --image_format jpg --annotation_format txt --xml2txt --stats
python main.py --root_dir /data/annotated_dataset --output_dir /data/annotated_dataset/annotations --image_format jpg --annotation_format txt --xml2txt --txt2json --train_ratio 0.8