https://github.com/apachecn-archive/igv
Science Score: 13.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
○.zenodo.json file
-
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (11.9%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: apachecn-archive
- Language: Python
- Default Branch: main
- Size: 78.1 KB
Statistics
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Invariant Grounding for Video Question Answering 🔥
Overview
This repo contains source code for Invariant Grounding for Video Question Answering (CVPR 2022 Oral, Best Paper Finalists). In this work, propose a new learning framework, Invariant Grounding for VideoQA (IGV), to ground the question-critical scene, whose causal relations with answers are invariant across different interventions on the complement. With IGV, the VideoQA models are forced to shield the answering process from the negative influence of spurious correlations, which significantly improves the reasoning ability.
<
Installation
- Main packages: PyTorch = 1.11
- See
requirements.txtfor other packages.
Data Preparation
We use MSVD-QA as an example to help get farmiliar with the code. Please download the dataset from here and the pre-computed features here
After downloading the data, please modify your data path and feature path in run.py.
Run IGV
Simply run run.sh to reproduce the results in the paper. We have saved our checkpoint here (acc 41.42% on MSVD-QA) for your references.
Reference
@InProceedings{Li_2022_CVPR,
author = {Li, Yicong and Wang, Xiang and Xiao, Junbin and Ji, Wei and Chua, Tat-Seng},
title = {Invariant Grounding for Video Question Answering},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2022},
pages = {2928-2937}
}
Acknowledgement
Our reproduction of the methods is based on the respective official repositories and NExT-QA, we thank the authors to release their code.
Owner
- Name: ApacheCN 归档
- Login: apachecn-archive
- Kind: organization
- Email: wizard.z@qq.com
- Repositories: 180
- Profile: https://github.com/apachecn-archive
防止重要项目丢失而设立的归档
GitHub Events
Total
Last Year
Dependencies
- Pillow ==8.2.0
- PyWavelets ==1.1.1
- PyYAML ==5.4.1
- Pygments ==2.9.0
- bert-score ==0.3.6
- block ==0.0.5
- block.bootstrap.pytorch ==0.1.6
- bootstrap.pytorch ==0.0.13
- certifi ==2020.12.5
- chardet ==4.0.0
- click ==8.0.0
- cycler ==0.10.0
- dataclasses ==0.8
- decorator ==4.4.2
- entrypoints ==0.3
- filelock ==3.3.1
- future ==0.18.2
- h5py ==2.7.1
- huggingface-hub ==0.0.19
- idna ==2.10
- imageio ==2.9.0
- importlib-metadata ==4.8.1
- importlib-resources ==5.1.3
- ipdb ==0.13.7
- jedi ==0.18.0
- joblib ==1.1.0
- kiwisolver ==1.3.1
- matplotlib ==3.3.4
- munch ==2.5.0
- networkx ==2.5.1
- nltk ==3.3
- numpy ==1.19.5
- opencv-python ==4.5.2.52
- packaging ==21.0
- pandas ==1.1.2
- plotly ==4.14.3
- pretrainedmodels ==0.7.4
- prompt-toolkit ==3.0.18
- protobuf ==3.17.0
- pyparsing ==2.4.7
- python-dateutil ==2.8.1
- pytz ==2021.1
- pywsd ==1.2.4
- regex ==2021.10.23
- requests ==2.25.1
- retrying ==1.3.3
- sacremoses ==0.0.46
- scikit-image ==0.17.2
- scipy ==1.5.4
- seaborn ==0.11.1
- skipthoughts ==0.0.1
- tabulate ==0.8.9
- tensorboardX ==2.2
- tifffile ==2020.9.3
- tokenizers ==0.10.3
- toml ==0.10.2
- torch ==1.6.0
- torchvision ==0.7.0
- tqdm ==4.60.0
- transformers ==4.11.3
- typing-extensions ==3.10.0.2
- urllib3 ==1.26.4
- wn ==0.0.23
- zipp ==3.4.1