Data

Tools

Indexes

Applications

Experiments

Open Source Science

ke-rcnn

Official implementation of KE-R-CNN for part-level attribute parsing.

https://github.com/josonchan/ke-rcnn

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (6.0%) to scientific vocabulary
Last synced: 10 months ago · JSON representation ·

Repository

Official implementation of KE-R-CNN for part-level attribute parsing.

Basic Info
  • Host: GitHub
  • Owner: JosonChan
  • License: apache-2.0
  • Language: Python
  • Default Branch: main
  • Size: 6.05 MB
Statistics
  • Stars: 8
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 4 years ago · Last pushed almost 4 years ago
Metadata Files
Readme License Citation

README.md

KE-R-CNN

Official implementation of KE-R-CNN for part-level attribute parsing. The new repository has been released at KE-RCNN, please follow the new code. This repository will not be maintained!

Installation

Dataset

You need to download the datasets and annotations follwing this repo's formate

Make sure to put the files as the following structure:

├─data │ ├─fashionpedia │ │ ├─train │ │ ├─test │ │ │─instances_attribute_train2020.json │ │ │─instances_attribute_val2020.json | | |─train_attr_knowledge_matrix.npy | ├─work_dirs | ├─KE-RCNN_r50_1x | | ├─latest.pth

Results and Models

FashionPedia

| Backbone | LR | APiou/APiou+f1 | APmaskiou/APmaskiou+f1 | DOWNLOAD | |--------------|:----:|:----------------:|:--------------------------:|:--------:| | R-50 | 1x | 41.9/39.1 | 37.5/36.2 |model| | R-101 | 1x | 43.8/39.9 | 38.2/36.0 |model| | Cascade-R-50| 1x | 44.0/41.0 | 37.5/36.5 |model| | Cascade-R101| 1x | 46.1/42.7 | 39.0/37.5 |[model] | | HRNet-w18 | 1x | 39.6/36.4 | -/- |[model] | | HRNet-w32 | 1x | 44.3/39.0 | -/- |[model] | | Swin-tiny | 1x | 44.3/42.1 | 40.6/38.6 |model| | Swin-small | 1x | 47.2/44.3 | 42.1/40.5 |[model] |

The effect of prior knowledge

| Backbone | LR | Fashionpedia/Wikipedia(AP_iou+f1) | DOWNLOAD | |--------------|:----:|:---------------------------------:|:--------:| | R-50 | 1x | 39.1/39.6 |[model] | | R-101 | 1x | 39.9/40.7 |[model] | | Cascade-R-50| 1x | 41.2/41.6 |[model] | | Cascade-R101| 1x | 42.7/42.3 |[model] | | HRNet-w18 | 1x | 36.4/37.7 |[model] | | HRNet-w32 | 1x | 39.0/39.2 |[model] | | Swin-tiny | 1x | 42.1/41.7 |[model] | | Swin-small | 1x | 44.3/45.0 |[model] |

Kinetics-TPS

| Backbone | LR | Accp | Accs |AP_part | DOWNLOAD | |--------------|:----:|:----------------:|:------------:|:-----------:|:--------:| | R-50 | 1x | 53.53 | 69.77 | 84.75 |[model] | | Cascade-R-50| 1x | 53.19 | 69.20 | 84.19 |[model] | | HRNet-w32 | 1x | 54.51 | 70.41 | 86.20 |[model] | | Swin-tiny | 1x | 56.20 | 72.24 | 86.77 |[model] | | Swin-small | 1x | 56.97 | 72.60 | 87.61 |[model] |

Evaluation

```

inference

CUDAVISIBLEDEVICES=0,1,2,3,4,5,6,7 ./tools/disttest.sh configs/KE-RCNN/KE-RCNNr501x.py workdirs/KE-RCNNr501x/latest.pth 8 --format-only --eval-options "jsonfileprefix=./KE-RCNNr501xval_result"

eval, noted that should change the json path produce by previous step.

python eval/fashion_eval.py ```

Training

Coming soon...

Citation

@article{KE-RCNN, Title = {KE-RCNN: Unifying Knowledge based Reasoning into Part-level Attribute Parsing}, Author = {Xuanhan Wang and Jingkuan Song and Xiaojia Chen and Lechao Cheng and Lianli Gao and Heng Tao Shen}, Year = {2022}, Eprint = {arXiv:2206.10146}, }

Owner

  • Name: JosonChan
  • Login: JosonChan
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.2.0
message: "If you use this software, please cite it as below."
authors:
  - name: "MMDetection Contributors"
title: "OpenMMLab Detection Toolbox and Benchmark"
date-released: 2018-08-22
url: "https://github.com/open-mmlab/mmdetection"
license: Apache-2.0

GitHub Events

Total
Last Year

Dependencies

docker/Dockerfile docker
  • pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
docker/serve/Dockerfile docker
  • pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
requirements/build.txt pypi
  • cython *
  • numpy *
requirements/docs.txt pypi
  • docutils ==0.16.0
  • recommonmark *
  • sphinx ==4.0.2
  • sphinx-copybutton *
  • sphinx_markdown_tables *
  • sphinx_rtd_theme ==0.5.2
requirements/mminstall.txt pypi
  • mmcv-full >=1.3.8
requirements/optional.txt pypi
  • cityscapesscripts *
  • imagecorruptions *
  • scipy *
  • sklearn *
requirements/readthedocs.txt pypi
  • mmcv *
  • torch *
  • torchvision *
requirements/runtime.txt pypi
  • matplotlib *
  • numpy *
  • pycocotools *
  • pycocotools-windows *
  • six *
  • terminaltables *
requirements/tests.txt pypi
  • asynctest * test
  • codecov * test
  • flake8 * test
  • interrogate * test
  • isort ==4.3.21 test
  • kwarray * test
  • mmtrack * test
  • onnx ==1.7.0 test
  • onnxruntime >=1.8.0 test
  • pytest * test
  • ubelt * test
  • xdoctest >=0.10.0 test
  • yapf * test
requirements.txt pypi
setup.py pypi