Link the dataset folder as 'data/', should contains 'image' and 'xml' for image and annotations respectively. pipeline create_labelmap.py create_splits.py create_name_size.py create_lmdb.sh ssd.py