search

stevenlsw / hoi-forecast

[CVPR 2022] Joint hand motion and interaction hotspots prediction from egocentric videos
MIT License
54 stars 9 forks source link

The "data.lmdb" file in data seem to be corrupted #16

Open Gggggjr opened 3 months ago

Gggggjr commented 3 months ago

Dear Stevenlsw, I hope this message finds you well. I encountered an issue that the "data.lmdb" file in data seem to be corrupted .I use python==3.6 and LMDB version is 1.2.1. This is my error report.

image
stevenlsw commented 3 months ago

hi, i think this is the keys in the training set, note in the provided lmdb we only store testing keys, the training keys needs to be generated follow the instructions. If you not sure how, please contact me and i could give you instructions.

Gggggjr commented 2 months ago

hi, I would be very grateful if you could give me a more detailed instructions.

JaehongJaehongMin commented 2 weeks ago

Thank you for kindly sharing the implementation.

However, I'm encountering the same issue. Could you please explain how to handle this key error?

JaehongJaehongMin commented 2 weeks ago

RULSTM repository explains how to download mdb files set consists of rgb, flow, obj mdb files. But I have no idea how to get lmdb file or convert mdb file to lmdb file.

stevenlsw commented 1 week ago

Hi, we only provide the key in testing split, for training split, you need to generate key pair by yourself as it take too much storage. For conversion, you could refer to this scripts and extract the training data yourself.