For extracting the provided features, did you resize the frames to a particular size

mever-team / distill-and-select

Authors official PyTorch implementation of the "DnS: Distill-and-Select for Efficient and Accurate Video Indexing and Retrieval" [IJCV 2022]

Apache License 2.0

64 stars 9 forks source link

For extracting the provided features, did you resize the frames to a particular size #9

Closed rliptech closed 1 year ago

rliptech commented 2 years ago

I tried running the feature_extractor, but the output size of region_vectors from layer3 is incompatible with region_vectors from other layers. My input is size [64, 3, 240, 320], region_vector of layer 1 has size[64, 256, 3, 4], layer2 [64, 512, 3, 4], layer 3 [64, 1024, 4, 5], layer4 [64, 2048, 3, 4]. Could you provide more information on the settings on how the extracted frames are preprocessed before extracting the resnet features? Thank you!

gkordo commented 1 year ago

Sorry for the late response. The issue has been fixed in f07be5f6a237eb5405e2630d8065487f7a930afd.