search

coldmanck / VidHOI

Official implementation of "ST-HOI: A Spatial-Temporal Baseline for Human-Object Interaction Detection in Videos" (ACM ICMRW 2021)
https://dl.acm.org/doi/10.1145/3463944.3469097
Apache License 2.0
50 stars 12 forks source link

Missing implementation of parse_bbox_file #19

Closed Jonas-Zhang97 closed 5 months ago

Jonas-Zhang97 commented 5 months ago

Hello, @coldmanck

Thank you for your great work.

I'm currently trying to train and run the st-hoi based on vidor, but it seems that one of the required module, i.e. parse_bbox_file, for demo (with input video or webcam) is not implemented in ava_helper, is there any way to get the module?

The error message looks like follows:

Traceback (most recent call last):
  File "tools/run_net_vidor.py", line 8, in <module>
    from demo_net_vidor import demo
  File "/home/zhihao/hoi_evaluation_project/hoi_models/hoi_st/tools/demo_net_vidor.py", line 11, in <module>
    from slowfast.visualization.ava_demo_precomputed_boxes import (
  File "/home/zhihao/hoi_evaluation_project/hoi_models/hoi_st/slowfast/visualization/ava_demo_precomputed_boxes.py", line 13, in <module>
    from slowfast.datasets.ava_helper import parse_bboxes_file
ImportError: cannot import name 'parse_bboxes_file'

Or alternatively is there any way to get the demo running?

Thank you in advance!

Jonas-Zhang97 commented 5 months ago

I see that the purpose of this module is to load "pre-computed predicted boxes" and thus maybe not needed for demo, so I will close this issue, but any comment is welcome