evaluation on my own dataset

[pfhil]

Hi, thank you for your work and for making it publicly available!
I would be really interested into trying out your evaluation method on my own dataset, in particular producing flow results (I am not really interested in metric evaluation not having ground truth flow) using your checkpoints. I have LIDAR data (as binary files), poses and also ground segmentation which I saw should be the only things needed for evaluation, but I wanted to ask a couple of hints on what I should be really modifying to load my dataset. It would be really helpful for me!

[Qingwen Zhang]

Hi,

If you want to evaluate your own dataset, you need to have flow ground truth available which may hard to get.

But if you only want to use pretrained weight to run on your own dataset, you can follow these lines and format your dataset to h5 file, then run vis.py file (more tutorial is on README):

DeFlow/dataprocess/extract_av2.py

Lines 192 to 201 in 6f0321a

 def create_group_data(group, pc, gm, pose, flow_0to1=None, flow_valid=None, flow_category=None, ego_motion=None): 
     group.create_dataset('lidar', data=pc.astype(np.float32)) 
     group.create_dataset('ground_mask', data=gm.astype(bool)) 
     group.create_dataset('pose', data=pose.astype(np.float32)) 
     if flow_0to1 is not None: 
         # ground truth flow information 
         group.create_dataset('flow', data=flow_0to1.astype(np.float32)) 
         group.create_dataset('flow_is_valid', data=flow_valid.astype(bool)) 
         group.create_dataset('flow_category_indices', data=flow_category.astype(np.uint8)) 
         group.create_dataset('ego_motion', data=ego_motion.astype(np.float32)) 

Let me know if you have further question.