- Python3
- OpenCV 3.4
- GCP with GPUs
- https://www.albert.cm/projects/viewpoint_3d_pose/
- http://vrai.dii.univpm.it/re-id-dataset
- https://www.albert.cm/projects/ram_person_id/
dataset/ - cut heads dataset
final-take2/
https://docs.google.com/presentation/d/14u4VbT3fY137V4mZnPOO8NkiWFpT-450tGD39dheFoE/edit?usp=sharing
python generate_tfrecord.py --csv_input=data/train.csv --output_path=data/train.record
- download and put training / testing data into depth-train/data (h5 files for depth-map, labels for train/test - 4 files)
- generate csv files using
generate_csv_labelsfunction from prepare-dataset.ipynb - generate tfrecord files from csvs using generate_tfrecord.py (see above)
- follow tutorial to prepare config files
python legacy/train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/ssd_mobilenet_v1_coco.configpython legacy/eval.py --logtostderr --pipeline_config_path=training/ssd_mobilenet_v1_coco.config --checkpoint_dir=training/ --eval_dir=eval/eval parameters: https://github.com/tensorflow/models/blob/abd504235f3c2eed891571d62f0a424e54a2dabc/research/object_detection/protos/eval.proto#L8
tensorboard --logdir=evalpython export_inference_graph.py --input_type image_tensor --pipeline_config_path=training/ssd_mobilenet_v1_coco.config --trained_checkpoint_prefix=training/model.ckpt-400 --output_directory output- locale problem
export LC_ALL="en_US.UTF-8"
export LC_CTYPE="en_US.UTF-8"
sudo dpkg-reconfigure locales