https://github.com/charlespikachu/yslao
ECCV2022 - You Should Look at All Objects
Science Score: 10.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
○codemeta.json file
-
○.zenodo.json file
-
○DOI references
-
✓Academic publication links
Links to: arxiv.org -
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (6.9%) to scientific vocabulary
Repository
ECCV2022 - You Should Look at All Objects
Basic Info
- Host: GitHub
- Owner: CharlesPikachu
- License: apache-2.0
- Language: Python
- Default Branch: main
- Size: 2.24 MB
Statistics
- Stars: 69
- Watchers: 6
- Forks: 2
- Open Issues: 8
- Releases: 0
Metadata Files
README.md
Introduction
The official repository for "You Should Look at All Objects".
Abstract
Feature pyramid network (FPN) is one of the key components for object detectors. However, there is a long-standing puzzle for researchers that the detection performance of large-scale objects are usually suppressed after introducing FPN. To this end, this paper first revisits FPN in the detection framework and reveals the nature of the success of FPN from the perspective of optimization. Then, we point out that the degraded performance of large-scale objects is due to the arising of improper back-propagation paths after integrating FPN. It makes each level of the backbone network only has the ability to look at the objects within a certain scale range. Based on these analysis, two feasible strategies are proposed to enable each level of the backbone to look at all objects in the FPN-based detection frameworks. Specifically, one is to introduce auxiliary objective functions to make each backbone level directly receive the back-propagation signals of various-scale objects during training. The other is to construct the feature pyramid in a more reasonable way to avoid the irrational back-propagation paths. Extensive experiments on the COCO benchmark validate the soundness of our analysis and the effectiveness of our methods. Without bells and whistles, we demonstrate that our method achieves solid improvements (more than 2%) on various detection frameworks: one-stage, two-stage, anchor-based, anchor-free and transformer-based detectors.
Owner
- Login: CharlesPikachu
- Kind: user
- Location: HongKong
- Company: HKU
- Website: https://charlespikachu.github.io/
- Repositories: 27
- Profile: https://github.com/CharlesPikachu
GitHub Events
Total
Last Year
Dependencies
- albumentations >=0.3.2
- cython *
- numpy *
- docutils ==0.16.0
- myst-parser *
- sphinx ==4.0.2
- sphinx-copybutton *
- sphinx_markdown_tables *
- sphinx_rtd_theme ==0.5.2
- mmcv-full >=1.3.17
- cityscapesscripts *
- imagecorruptions *
- scipy *
- sklearn *
- timm *
- mmcv *
- torch *
- torchvision *
- matplotlib *
- numpy *
- pycocotools *
- six *
- terminaltables *
- asynctest * test
- codecov * test
- flake8 * test
- interrogate * test
- isort ==4.3.21 test
- kwarray * test
- onnx ==1.7.0 test
- onnxruntime >=1.8.0 test
- protobuf <=3.20.1 test
- pytest * test
- ubelt * test
- xdoctest >=0.10.0 test
- yapf * test