Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 762 Bytes

README.md

File metadata and controls

20 lines (14 loc) · 762 Bytes

non-offical implementation of undeepvo

  1. Prepare dataset run generate_train.m with matlab

  2. Activate tensorflow source ~/tensorflow/bin/activate

  3. Training python undeepvo_main.py --mode train --model_name my_model --data_path /media/youngji/storagedevice/naver_data/kitti_odometry/dataset/ --filenames_file ./utils/filenames/kitti_train_files2.txt --log_directory ./tmp/

  4. visualization tensorboard --logdir=./tmp/my_model

  5. Testing python monodepth_main.py --mode test --data_path /media/youngji/storagedevice/naver_data/kitti_odometry/dataset/
    --filenames_file ./utils/filenames/kitti_train_files2.txt --log_directory ./tmp/
    --checkpoint_path ./tmp/my_model/model-181250

python undeepvo_test.py --checkpoint_path ./tmp/my_model3/model-30000