search

wenbowen123 / iros20-6d-pose-tracking

[IROS 2020] se(3)-TrackNet: Data-driven 6D Pose Tracking by Calibrating Image Residuals in Synthetic Domains
Other
384 stars 66 forks source link

predict_ros.py error finding model path #71

Open ifeeney opened 7 months ago

ifeeney commented 7 months ago

Hi,

I was trying to run predict_ros.py and noticed that it does not pass a model_path into the Tracker() constructor. Looking at past commits, I see you made a change to predict.py to default the model_path argument to none. I think that change may be causing this error when running predict_ros.py... /home/se3_tracknet/pose_init.txt Traceback (most recent call last): File "predict_ros.py", line 109, in <module> tracker = Tracker(dataset_info, images_mean, images_std,ckpt_dir,trans_normalizer=dataset_info['max_translation'],rot_normalizer=dataset_info['max_rotation']) File "/home/se3_tracknet/predict.py", line 131, in __init__ mesh = trimesh.load(model_path) File "/usr/local/lib/python3.6/dist-packages/trimesh/exchange/load.py", line 159, in load file_type) ValueError: File type: nonetype not supported

wenbowen123 commented 7 months ago

Thanks for bringing this up! Like you said, it is possible that the change for the predict.py did not take into account the predict_ros.py. First of all, did you successfully run the predict.py for the public datasets?

ifeeney commented 7 months ago

I was able to successfully run predict.py with the example data but ran into an issue when running predict_ros.py. I was able to narrow down the problem to that particular change.

wenbowen123 commented 7 months ago

Great to hear that you got success running the examples. For ros, you might need a bit modification, such as providing model_path which is the path to the CAD model (refer to the example)