Some utils to convert the ecostack pollinator detection dataset (available here: https://vision.eng.au.dk/ecostack/) from YOLO annotation format into MSCOCO, visualwakewords (vww, see: https://arxiv.org/abs/1906.05721 and https://blog.tensorflow.org/2019/10/visual-wake-words-with-tensorflow-lite_30.html). A TFRecord conversion is also required to train using repo accessible here: https://github.com/rossGardiner/visual-wake-words
Files dataset_utils.py
and download_and_convert_visualwakewords_lib.py
are copied from tf_slim library. See: https://github.com/tensorflow/models/blob/master/research/slim/README.md
wget https://vision.eng.au.dk/?download=/data/EcoStack/insects/train1201.zip
wget https://vision.eng.au.dk/?download=/data/EcoStack/insects/val1201.zip
wget https://vision.eng.au.dk/?download=/data/EcoStack/insects/test1201.zip
unzip *.zip
Should leave three directories: train1201/
, test1201/
, val1201/
. You may need to rename stuff
python3 -m virtualenv venv
source venv/bin/activate
pip3 install -r requirements.txt
NB: not a lean requirements.txt
Check the script args:
python3 labels_convert_yolo_mscoco.py --help
usage: labels_convert_yolo_mscoco.py [-h] [-b BINARY] [-p PREFIX] [-a ANNOTATION_DIR]
optional arguments:
-h, --help show this help message and exit
-b BINARY, --binary BINARY
binary classes or not, 1 or 0 (yes or no)
-p PREFIX, --prefix PREFIX
prefix, train, test or val
-a ANNOTATION_DIR, --annotation_dir ANNOTATION_DIR
annotation directory
python3 labels_convert_yolo_mscoco.py -p train
python3 labels_convert_yolo_mscoco.py -p val
python3 labels_convert_yolo_mscoco.py -p test
python3 labels_convert_mscoco_vww.py -p train
python3 labels_convert_mscoco_vww.py -p val
python3 labels_convert_mscoco_vww.py -p test
python3 labels_convert_vww_tfrecord.py -p train
python3 labels_convert_vww_tfrecord.py -p val -s 10
python3 labels_convert_vww_tfrecord.py -p test
Play around with inspect_tfrecord.py
to check the contents of a resultant tfrecord file