avd-anonymizer
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 (6.9%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: wenjie-hoo
- Language: Jupyter Notebook
- Default Branch: main
- Size: 213 MB
Statistics
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Optimizing privacy-aware object detection on the PP4AV benchmark with CrossKD + Uncertainty-Weighted KD (UWKD)
Overview
CrossKD tackles the target‑conflict problem of dense detectors by routing student feature maps through a frozen teacher head. In this project, we extend CrossKD with an Uncertainty‑Weighted KD loss (UWKD) that down‑weights ambiguous teacher logits.
Install
```bash
1. clone
$ git clone https://github.com/your‑user/CrossKD‑PP4AV.git $ cd CrossKD
2. create conda env
$ conda create --name mmdet python=3.8 -y $ conda activate mmdet
3. install pytorch
$ conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
4. build MMDetection & ops
$ pip install -U openmim $ mim install "mmengine==0.7.3" $ mim install "mmcv==2.0.0rc4" ```
Training
Train teacher (optional)
bash
$ python tools/train.py configs/fcos/${CONFIG_FILE} [optional arguments]
Pre‑trained checkpoints are also available here or
bash
$ pip install gdown
$ gdown https://drive.google.com/uc?id=1vMo2Oflzm7nnw18rHPJqk-9gPPG4NEz6
Distill student
bash
$ python tools/train.py configs/crosskd+uwkd/${CONFIG_FILE} [optional arguments]
Evaluate
coco style metrics
bash
$ python tools/test.py configs/crosskd+uwkd/${CONFIG_FILE} ${CHECKPOINT_FILE}
mmdet benchmark
bash
$ python tools/analysis_tools/benchmark.py configs/crosskd+uwkd/${CONFIG_FILE} --checkpoint ${CHECKPOINT_FILE} [optional arguments]
Comparison Table
Dataset
PP4AV dataset: A collection of images and videos captured in various urban environments, annotated with bounding boxes for pedestrian faces and license plates.
Acknowledgements
- MMDetection and the OpenMMLab team
- CrossKD authors for the original implementation
Owner
- Name: Wenjie Hu
- Login: wenjie-hoo
- Kind: user
- Location: Dublin
- Repositories: 7
- Profile: https://github.com/wenjie-hoo
Ciao! :bowtie:
Citation (CITATION.cff)
cff-version: 1.2.0
message: "If you use this software, please cite it as below."
title: "AVD-Anonymizer"
version: "1.0.0"
date-released: 2025-07-08
authors:
- family-names: Hu
given-names: Wenjie
repository-code: https://github.com/wenjie-hoo/AVD-Anonymizer
GitHub Events
Total
- Member event: 1
- Push event: 21
- Create event: 1
Last Year
- Member event: 1
- Push event: 21
- Create event: 1
Dependencies
- Pillow >=7.1.2
- PyYAML >=5.3.1
- ipython *
- matplotlib >=3.2.2
- numpy >=1.18.5,<1.24.0
- opencv-python >=4.1.1
- pandas >=1.1.4
- protobuf <4.21.3
- psutil *
- requests >=2.23.0
- scipy >=1.4.1
- seaborn >=0.11.0
- tensorboard >=2.4.1
- thop *
- torch >=1.7.0,
- torchvision >=0.8.1,
- tqdm >=4.41.0
- pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
- pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
- pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
- pytorch/pytorch ${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel build
- pytorch/pytorch 2.0.0-cuda11.7-cudnn8-runtime build
- gcr.io/google-appengine/python latest build
- 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 >=2.0.0rc4,<2.1.0
- mmengine >=0.4.0,<1.0.0
- cityscapesscripts *
- imagecorruptions *
- scikit-learn *
- mmcv >=2.0.0rc1,<2.1.0
- mmengine >=0.1.0,<1.0.0
- scipy *
- torch *
- torchvision *
- matplotlib *
- numpy *
- pycocotools *
- scipy *
- six *
- terminaltables *
- asynctest * test
- cityscapesscripts * test
- codecov * test
- flake8 * test
- imagecorruptions * test
- instaboostfast * test
- interrogate * test
- isort ==4.3.21 test
- kwarray * test
- memory_profiler * test
- onnx ==1.7.0 test
- onnxruntime >=1.8.0 test
- parameterized * test
- protobuf <=3.20.1 test
- psutil * test
- pytest * test
- ubelt * test
- xdoctest >=0.10.0 test
- yapf * test
- matplotlib >=3.3.0
- numpy >=1.22.2
- opencv-python >=4.6.0
- pandas >=1.1.4
- pillow >=7.1.2
- psutil *
- py-cpuinfo *
- pyyaml >=5.3.1
- requests >=2.23.0
- scipy >=1.4.1
- seaborn >=0.11.0
- thop >=0.1.1
- torch >=1.8.0
- torchvision >=0.9.0
- tqdm >=4.64.0
- ultralytics >=8.1.47
- PyYAML >=5.3.1
- gdown >=5.2.0
- gitpython >=3.1.30
- matplotlib >=3.3
- numpy >=1.23.5
- opencv-python >=4.1.1
- pandas >=1.1.4
- pillow >=10.3.0
- psutil *
- requests >=2.32.2
- scipy >=1.4.1
- seaborn >=0.11.0
- setuptools >=70.0.0
- thop >=0.1.1
- torchvision >=0.9.0
- tqdm >=4.66.3
- Flask ==2.3.2
- gunicorn ==22.0.0
- pip ==23.3
- werkzeug >=3.0.1
- zipp >=3.19.1