giorking/Dataset_to_VOC_converter

Scripts to convert datasets (Caltech pedestrian, MS COCO, HDA) to PASCAL VOC format

These scripts are used for convert datasets (MS COCO, Caltech pedestrian dataset) to PASCAL VOC format for later training.

Requirements

• MS COCO toolbox
• cytoolz
• lmxl
• scipy, numpy

Usage

anno_json_image_urls.py: extract image url (coco source not filckr) from annotation json file. See anno_json_image_urls.sh
download_coco_images.py: download coco image files from given urls (extracted from instance/keypoint annotation json file) . See download.sh
anno_coco2voc.py: convert coco annotation json file to VOC xml files. See anno_coco2voc.sh
vbb2voc.py: extract images with person bbox in seq file and convert vbb annotation file to xml files.