search

wjun0830 / QD-DETR

Official pytorch repository for "QD-DETR : Query-Dependent Video Representation for Moment Retrieval and Highlight Detection" (CVPR 2023 Paper)
https://arxiv.org/abs/2303.13874
Other
183 stars 13 forks source link

about feature extract #20

Closed hhhhnwl closed 12 months ago

hhhhnwl commented 1 year ago

1.To use the pretrain model, could you please share the script for slowfast feature extraction? 2.Are the parameters for extracting CLIP features exactly the same as the moment-detr?

hhhhnwl commented 1 year ago

clip_len maybe set to 1? python extract_feature/extract.py --dataflow --csv /output/csv/slowfast_info.csv \ --batch_size 45 --num_decoding_thread 4 --clip_len 1 \ TEST.CHECKPOINT_FILE_PATH /models/SLOWFAST_8x8_R50.pkl

wjun0830 commented 1 year ago

Hello Thanks for the interest.

  1. For the feature extraction, you can refer to HERO github.
  2. Yes, we followed the procedure from our baseline, Moment-DETR.
  3. Yes, we set clip len to 1.