Bilinear3D for Paired Egocentric Interaction Recognition

This repo is the implementation for the paper (https://arxiv.org/abs/2003.10663).

Overview

This code is based on pytorch-i3d

This code was tested with PyTorch 1.0

Directories shall be organized as : data_root/video_name/img_00001.jpg rgb

data_root/video_name/flow_x_00001.jpg optical flow x

data_root/video_name/flow_y_00001.jpg optical flow y

The split file follows the format: video_name label

run train_i3d.sh, please refer to comments in pytorch_i3d.py for arguments.

Name		Name	Last commit message	Last commit date
Latest commit History 78 Commits
data		data
.gitignore		.gitignore
BPC3D.py		BPC3D.py
BPI3D.py		BPI3D.py
BPTSN.py		BPTSN.py
BilinearPooling.py		BilinearPooling.py
C3D.py		C3D.py
CompactBilinearPooling.py		CompactBilinearPooling.py
CompactBilinearPoolingFourStream.py		CompactBilinearPoolingFourStream.py
CorrelogramPooling.py		CorrelogramPooling.py
DI3D.py		DI3D.py
DWT.py		DWT.py
IG.py		IG.py
LICENSE.txt		LICENSE.txt
R2Plus1D.py		R2Plus1D.py
README.md		README.md
W3D.py		W3D.py
attack.py		attack.py
charades_dataset.py		charades_dataset.py
charades_dataset_full.py		charades_dataset_full.py
confusion_matrix_figure.py		confusion_matrix_figure.py
extract_features.py		extract_features.py
feature_visualize.py		feature_visualize.py
inceptionv3.py		inceptionv3.py
mi3d.py		mi3d.py
multi_label_loss.py		multi_label_loss.py
pev.py		pev.py
prediction.py		prediction.py
prediction.sh		prediction.sh
pytorch_i3d.py		pytorch_i3d.py
result.md		result.md
scores_to_libsvm.py		scores_to_libsvm.py
spatial_transform.py		spatial_transform.py
target_transform.py		target_transform.py
temporal_transform.py		temporal_transform.py
test.py		test.py
train.sh		train.sh
train_i3d.py		train_i3d.py
train_i3d.sh		train_i3d.sh
tsn.py		tsn.py
ucf.py		ucf.py
videotransforms.py		videotransforms.py