search

yzcjtr / GeoNet

Code for GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018)
MIT License
726 stars 181 forks source link

Output problem #56

Closed TopGun666 closed 5 years ago

TopGun666 commented 5 years ago

Hi @yzcjtr

Thanks for your work. I have 2 questions about the code.

  1. When I run the eval_pose.py, I can't get any output of the evaluation. 2.When I run the test_depth, I only get the .npy file, but it can't be loaded by evaluate code.

Kind regards,

Yu

yzcjtr commented 5 years ago

Hi, can you post your log file here for further inspection? Have you configured all the arguments correctly?

TopGun666 commented 5 years ago

Thanks for your reply. The log file is a little big (about 148 MB). Could I sent the screenshot to your email? Because the picture can't be updated here. How about the question 1 ? I can't get any pose evaluation result.

yzcjtr commented 5 years ago

I meant the error log printed to your screen while running the command. You can also post your script here.

TopGun666 commented 5 years ago

There is no error log. It runs normally. When I run the first 2 scripts, It ouput is normal. But I can't get any output of the evaluation. Here is my script.

python geonet_main.py --mode=test_pose --dataset_dir=/home/yhe3/Data/dataset/kitti_odom/ --init_ckpt_file=/home/yhe3/Data/pretrained_models/GeoNet/models/geonet_posenet/model --batch_size=1 --seq_length=5 --pose_test_seq=9 --output_dir=/home/yhe3/Data/project/GeoNet/predictions/poses/09/

python kitti_eval/generate_pose_snippets.py --dataset_dir=/home/yhe3/Data/dataset/kitti_odom/ --output_dir=/home/yhe3/Data/project/GeoNet/predictions/poses/snippets/09/ --seq_id=09 --seq_length=5

python kitti_eval/eval_pose.py --gtruth_dir=/home/yhe3/Data/project/GeoNet/predictions/poses/snippets/09/ --pred_dir=/home/yhe3/Data/project/GeoNet/predictions/poses/09/

yzcjtr commented 5 years ago

The script looks good to me. Can you confirm the content of the generated pose snippets and predictions? Are they normal? Is your python environment consistent with ours? In addition, try "python -u" instead of "python".