Unified Change Detection Framework

Powered by Huggingface Hub 🤗

Contributors:

Weikang Yu, Xiaokang Zhang, Richard Gloaguen, Xiao Xiang Zhu, Pedram Ghamisi

News

11.11.2024 UCD is open to everyone! Be a contributor by sending a pull request!

11.11.2024 Codes for UCD have been released, if you find any problems or bugs, please leave us a message.

09.11.2024 Our paper of MineNetCD has been published on IEEE TGRS 2024, the repo for MineNetCD is available here.

09.07.2024 The UCD project is announced on IEEE IGARSS 2024, we are organizing the codes.

Environment Preparation:

Create a conda environment for UCD

conda create -n ucd python=3.10
conda activate ucd
conda install pytorch torchvision pytorch-cuda=12.1 -c pytorch -c nvidia
pip install -r requirements.txt

Configurate the accelerate package:

accelerate config

How to use:

To train a model:

accelerate launch train.py --config $CONFIG$

An example of CONFIG can be configs/DTCDSCN_MNCD256.yml

Or if you want to run the framework in Singularity & Slurm, you can use this command:

srun singularity exec --env PREPEND_PATH=/home/yu34/.local/bin --nv /home/yu34/ucd.sif accelerate launch train.py --config $CONFIG$

To test a model:

accelerate launch test.py --model $PRETRAINED_MODEL_PATH$

An example of PRETRAINED_MODEL_PATH can be a local path like checkpoints/MNCD256/ResUnet/BestF1/ or a Huggingface hub id like HZDR-FWGEL/UCD-MNCD256-ResUnet

To push a pretrained model to Huggingface hub in the UCD format:

accelerate launch test.py --model $LOCAL_PATH$ --batch-size 10 --push-to-hub $HUB_NAME$

The example of LOCAL_PATH can be checkpoints/MNCD256/ResUnet/BestF1/, the example of HUB_NAME can be HZDR-FWGEL/UCD-MNCD256-ResUnet

To transfer a model from torch format (e.g., pytorch_model.bin) to UCD compatible format:

You should prepare a config file in your directory, an example can be found via configs/MineNetCD_MNCD256.yml

accelerate launch test.py --external-config configs/MineNetCD_MNCD256.yml  --batch-size 10 --push-to-hub HZDR-FWGEL/UCD-MNCD256-ChangeFFT

The use_external_checkpint will be set to false before uploading the model to the hub.

To calculate model parameters & FLOPs:

python params_sum.py --config $CONFIG$

Available Models:

Model	Backbone	#Params	FLOPs	Source
A2Net	VGG16	3.60M	2.86G	IEEE TGRS 2023
AFCF3D	ResNet18	16.83M	29.54G	IEEE TGRS 2023
BIT	ResNet18	11.39M	8.28G	IEEE TGRS 2021
CGNet	VGG16	37.18M	81.66G	IEEE JSTARS 2023
ChangeFormer	MIT	39.13M	129.7G	IEEE IGARSS 2022
DMINet	ResNet18	6.44M	16.16G	IEEE TGRS 2021
DTCDSCN	-	39.17M	10.72G	IEEE GRSL 2020
FC-EF	-	1.29M	2.92G	IEEE ICIP 2018
FCNPP	-	14.56M	43.10G	IEEE GRSL 2019
HCGMNet	VGG16	45.13M	301.82G	IEEE IGARSS 2023
ICIFNet	ResNet18	24.64M	22.97G	IEEE TGRS 2022
MSPSNet	-	2.11M	13.89G	IEEE TGRS 2021
RDPNet	-	1.62M	1.63G	IEEE TGRS 2022
ResUnet	-	12.59M	28.51G	IEEE IGARSS 2021
SiamUnet-Conc	-	1.47M	4.55G	IEEE ICIP 2018
SiamUnet-Diff	-	1.29M	3.99G	IEEE ICIP 2018
SNUNet	-	11.48M	43.6G	IEEE GRSL 2021
TFI-GR	ResNet18	27.06M	9.09G	IEEE TGRS 2022
TinyCD	EfficientNet-b4	0.27M	1.44G	NCA 2023
MineNetCD	SwinT-Tiny	57.81M	63.28G	This Paper

Available Datasets:

Dataset	#Patches	Scenario	Location	Sensor	Resolution
CLCD	2400	Cropland	Guangdong, China	Gaofen-2	0.5m-2m
EGY-BCD	6091	Building	Egypt	Google Earth	0.25m
GVLM-CD	7496	Landslide	Global	Google Earth	0.59m
LEVIR-CD	10192	Building	Texas, USA	Google Earth	0.5m
SYSU-CD	20000	Urban	Hong Kong, China	Aerial Image	0.5m
MineNetCD	71711	Mining	Global	Google Earth	1.2m

Our Results:

Here are results derived from the UCD

CLCD256:

Dataset for these implementations: ericyu/CLCD_Cropped_256

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	CLCD256	0.9199	0.3765	0.4474	0.3250	0.2319	`HZDR-FWGEL/UCD-CLCD256-A2Net`
BIT	CLCD256	0.9488	0.6590	0.6526	0.6657	0.4915	`HZDR-FWGEL/UCD-CLCD256-BIT`
DMINet	CLCD256	0.9392	0.5744	0.5990	0.5517	0.4029	`HZDR-FWGEL/UCD-CLCD256-DMINet`
ICIFNet	CLCD256	0.9416	0.5629	0.6355	0.5052	0.3917	`HZDR-FWGEL/UCD-CLCD256-ICIFNet`
RDPNet	CLCD256	0.9288	0.5431	0.5194	0.569	0.3727	`HZDR-FWGEL/UCD-CLCD256-RDPNet`
SiamUNet-Diff	CLCD256	0.9358	0.4914	0.5983	0.4169	0.3257	`HZDR-FWGEL/UCD-CLCD256-SiamUDiff`
ChangeFormer	CLCD256	0.9431	0.6214	0.6151	0.6279	0.4508	`HZDR-FWGEL/UCD-CLCD256-ChangeFormer`

GVLM256:

Dataset for these implementations: ericyu/GVLM_Cropped256

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	GVLM256	0.9776	0.8114	0.9156	0.7285	0.6827	`HZDR-FWGEL/UCD-GVLM256-A2Net`
BIT	GVLM256	0.9841	0.8768	0.8974	0.8572	0.7807	`HZDR-FWGEL/UCD-GVLM256-BIT`
DMINet	GVLM2256	0.9825	0.8664	0.8738	0.8591	0.7643	`HZDR-FWGEL/UCD-GVLM256-DMINet`
ICIFNet	GVLM256	0.9831	0.8722	0.8735	0.871	0.7734	`HZDR-FWGEL/UCD-GVLM256-ICIFNet`
RDPNet	GVLM256	0.9827	0.868	0.875	0.8611	0.7668	`HZDR-FWGEL/UCD-GVLM256-RDPNet`
SiamUNet-Diff	GVLM256	0.9801	0.8431	0.8791	0.81	0.7288	`HZDR-FWGEL/UCD-GVLM256-SiamUDiff`
ChangeFormer	GVLM256	0.9831	0.8685	0.8943	0.8441	0.7675	`HZDR-FWGEL/UCD-GVLM256-ChangeFormer`

EGYBCD:

Dataset for these implementations: ericyu/EGY_BCD

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	EGY_BCD	0.9624	0.6914	0.7283	0.6581	0.5284	`HZDR-FWGEL/UCD-EGYBCD-A2Net`
BIT	EGYBCD	0.9735	0.7906	0.8016	0.7799	0.6537	`HZDR-FWGEL/UCD-EGYBCD-BIT`
DMINet	EGYBCD	0.9585	0.6929	0.6591	0.7304	0.5301	`HZDR-FWGEL/UCD-EGYBCD-DMINet`
ICIFNet	EGYBCD	0.9621	0.6903	0.7241	0.6595	0.5270	`HZDR-FWGEL/UCD-EGYBCD-ICIFNet`
RDPNet	EGYBCD	0.9612	0.6859	0.7125	0.6612	0.5220	`HZDR-FWGEL/UCD-EGYBCD-RDPNet`
SiamUNet-Diff	EGYBCD	0.9524	0.6422	0.6191	0.6671	0.4729	`HZDR-FWGEL/UCD-EGYBCD-SiamUDiff`
ChangeFormer	EGYBCD	0.9651	0.7181	0.7436	0.6944	0.5602	`HZDR-FWGEL/UCD-EGYBCD-ChangeFormer`

LEVIRCD256:

Dataset for these implementations: ericyu/LEVIRCD_Cropped256

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	LEVIRCD256	0.9699	0.6687	0.7613	0.5962	0.5023	`HZDR-FWGEL/UCD-LEVIRCD256-A2Net`
BIT	LEVIRCD256	0.9888	0.8884	0.9046	0.8728	0.7992	`HZDR-FWGEL/UCD-LEVIRCD256-BIT`
DMINet	LEVIRCD256	0.9845	0.8431	0.8708	0.8171	0.7287	`HZDR-FWGEL/UCD-LEVIRCD256-DMINet`
ICIFNet	LEVIRCD256	0.9827	0.8162	0.8871	0.7558	0.6895	`HZDR-FWGEL/UCD-LEVIRCD256-ICIFNet`
RDPNet	LEVIRCD256	0.9808	0.8058	0.8315	0.7816	0.6747	`HZDR-FWGEL/UCD-LEVIRCD256-RDPNet`
SiamUNet-Diff	LEVIRCD256	0.9805	0.5991	0.7822	0.6874	0.6423	`HZDR-FWGEL/UCD-LEVIRCD256-SiamUDiff`
ChangeFormer	LEVIRCD256	0.9826	0.8232	0.8516	0.7967	0.6996	`HZDR-FWGEL/UCD-LEVIRCD256-ChangeFormer`

SYSUCD:

Dataset for these implementations: ericyu/SYSU_CD

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	SYSUCD	0.8812	0.7598	0.7260	0.7969	0.6126	`HZDR-FWGEL/UCD-SYSUCD-A2Net`
BIT	SYSUCD	0.873	0.7497	0.7004	0.8064	0.5996	`HZDR-FWGEL/UCD-SYSUCD-BIT`
DMINet	SYSUCD	0.8881	0.7464	0.8014	0.6984	0.5954	`HZDR-FWGEL/UCD-SYSUCD-DMINet`
ICIFNet	SYSUCD	0.8640	0.703	0.7248	0.6825	0.5421	`HZDR-FWGEL/UCD-SYSUCD-ICIFNet`
RDPNet	SYSUCD	0.8852	0.7536	0.763	0.7445	0.6047	`HZDR-FWGEL/UCD-SYSUCD-RDPNet`
SiamUNet-Diff	SYSUCD	0.8546	0.5991	0.8563	0.4608	0.4277	`HZDR-FWGEL/UCD-SYSUCD-SiamUDiff`
ChangeFormer	SYSUCD	0.8912	0.7593	0.7938	0.7277	0.612	`HZDR-FWGEL/UCD-SYSUCD-ChangeFormer`

MineNetCD256:

Dataset for these implementations: HZDR-FWGEL/MineNetCD256

Model	Dataset	Accuracy	mF1	Precision	Recall	cIoU	Pretrained_Path
A2Net	MineNetCD256	0.9185	0.6404	0.7215	0.5758	0.4710	`HZDR-FWGEL/UCD-MNCD256-A2Net`
AFCF3D	MineNetCD256	0.8932	0.5772	0.5755	0.5789	0.4061	`HZDR-FWGEL/UCD-MNCD256-AFCF3D`*
BIT	MineNetCD256	0.9115	0.6227	0.6727	0.5795	0.4521	`HZDR-FWGEL/UCD-MNCD256-BIT`
ChangeFormer	MineNetCD256	0.8699	0.4995	0.4848	0.5151	0.3329	`HZDR-FWGEL/UCD-MNCD256-ChangeFormer`
CGNet	MineNetCD256	0.9004	0.547	0.6412	0.477	0.3765	`HZDR-FWGEL/UCD-MNCD256-CGNet`
DMINet	MineNetCD256	0.8963	0.5169	0.6257	0.4403	0.3485	`HZDR-FWGEL/UCD-MNCD256-DMINet`
DTCDSCN	MineNetCD256	0.8984	0.5567	0.6184	0.5068	0.3864	`HZDR-FWGEL/UCD-MNCD256-DTCDSCN`*
FC-EF	MineNetCD256	0.8836	0.415	0.5625	0.329	0.2619	`HZDR-FWGEL/UCD-MNCD256-FCEF`*
FCNPP	MineNetCD256	0.8549	0.3449	0.4004	0.3030	0.2084	`HZDR-FWGEL/UCD-MNCD256-FCNPP`
HCGMNet	MineNetCD256	0.9076	0.5876	0.6718	0.5222	0.4161	`HZDR-FWGEL/UCD-MNCD256-HCGMNet`
ICIFNet	MineNetCD256	0.8915	0.5018	0.5958	0.4334	0.3349	`HZDR-FWGEL/UCD-MNCD256-ICIFNet`
ChangeFFT	MineNetCD256	0.9251	0.6963	0.7120	0.6814	0.5343	`HZDR-FWGEL/UCD-MNCD256-ChangeFFT`
MSPSNet	MineNetCD256	0.8998	0.5591	0.6277	0.5041	0.388	`HZDR-FWGEL/UCD-MNCD256-MSPSNet`
RDPNet	MineNetCD256	0.8768	0.4961	0.5120	0.4811	0.3298	`HZDR-FWGEL/UCD-MNCD256-RDPNet`
ResUnet	MineNetCD256	0.8663	0.5072	0.4727	0.5488	0.3398	`HZDR-FWGEL/UCD-MNCD256-ResUnet`
SiamUNet-Conc	MineNetCD256	0.8979	0.5099	0.6460	0.4211	0.3422	`HZDR-FWGEL/UCD-MNCD256-SiamUConc`
SiamUNet-Diff	MineNetCD256	0.8956	0.3736	0.7671	0.2469	0.2297	`HZDR-FWGEL/UCD-MNCD256-SiamUDiff`
SNUNet	MineNetCD256	0.8988	0.5371	0.6351	0.4654	0.3675	`HZDR-FWGEL/UCD-MNCD256-SNUNet`*
TFI_GR	MineNetCD256	0.8932	0.5772	0.5755	0.5789	0.4061	`HZDR-FWGEL/UCD-MNCD256-TFIGR`*
TinyCD	MineNetCD256	0.8999	0.5648	0.625	0.5153	0.3936	`HZDR-FWGEL/UCD-MNCD256-TinyCD`

* Pretrained Models may only be loaded using accelerator with multiple graphical cards.

Tutorial Avaiable!

We just added a very simple example as a tutorial for those who are interested in change detection, check here for more details.

Future Development Schedule:

We will implement more models and datasets. If you are interested in this project and want to make any contributions, please send a pull request and we will add your names under the contributors!

If you have any questions or meeting any difficulties when using this framework, please leave us with an issue or you can contact us with email address: [email protected]

Acknowledgement:

We would like to thank Huggingface for providing a wonderful open-source platform. We would also like to thank all the authors and contributors who open-sourced the datasets and models that we incorporated into the UCD platform.

Citation

If you find MineNetCD useful for your study, please kindly cite us:

@ARTICLE{10744421,
  author={Yu, Weikang and Zhang, Xiaokang and Gloaguen, Richard and Zhu, Xiao Xiang and Ghamisi, Pedram},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={MineNetCD: A Benchmark for Global Mining Change Detection on Remote Sensing Imagery}, 
  year={2024},
  volume={},
  number={},
  pages={1-1},
  keywords={Data mining;Remote sensing;Feature extraction;Benchmark testing;Earth;Transformers;Annotations;Graphical models;Distribution functions;Sustainable development;Mining change detection;remote sensing;benchmark;frequency domain learning;unified framework},
  doi={10.1109/TGRS.2024.3491715}}

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
configs		configs
models		models
utils		utils
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
dataset_maker.py		dataset_maker.py
model.py		model.py
params_sum.py		params_sum.py
requirements.txt		requirements.txt
test.py		test.py
train.py		train.py

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Unified Change Detection Framework

Powered by Huggingface Hub 🤗

Contributors:

News

Environment Preparation:

How to use:

To train a model:

To test a model:

To push a pretrained model to Huggingface hub in the UCD format:

To calculate model parameters & FLOPs:

Available Models:

Available Datasets:

Our Results:

CLCD256:

GVLM256:

EGYBCD:

LEVIRCD256:

SYSUCD:

MineNetCD256:

Tutorial Avaiable!

Future Development Schedule:

Acknowledgement:

Citation

About

Releases

Packages

Languages

License

AI4RS/UCD

Folders and files

Latest commit

History

Repository files navigation

Unified Change Detection Framework

Powered by Huggingface Hub 🤗

Contributors:

News

Environment Preparation:

How to use:

To train a model:

To test a model:

To push a pretrained model to Huggingface hub in the UCD format:

To calculate model parameters & FLOPs:

Available Models:

Available Datasets:

Our Results:

CLCD256:

GVLM256:

EGYBCD:

LEVIRCD256:

SYSUCD:

MineNetCD256:

Tutorial Avaiable!

Future Development Schedule:

Acknowledgement:

Citation

About

Resources

License

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages