
<div align="center">

<h2 style="border-bottom: 1px solid lightgray;">Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion</h2>















<br/>

</div>

<!-- 
<img src="bs=16_test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->
<!-- 
<img src="test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->

<!-- As the training epochs increases, the test set accuracy of different methods. (Top: batchsize is 16. Bottom: batchsize is 1024) -->

<!-- 
<img src="temporal_analysis.png" alt="Framework" style="max-width: 90%; height: auto;"/>
Examples of growing window image reconstruction with 5 different random seeds. -->


<img src="fig-framework.png" alt="Framework" style="max-width: 100%; height: auto;"/>
Framework of our proposed method.




<!--  -->
<img src="fig-genexample.png" alt="fig-genexample" style="max-width: 90%; height: auto;"/>  

Some examples of using EEG to reconstruct stimulus images.


## News:
- [2024/09/26] Our paper is accepted to **NeurIPS 2024**.
- [2024/09/25] We have updated the [arxiv](https://arxiv.org/abs/2403.07721) paper.
- [2024/08/01] Update scripts for training and inference in different tasks.
- [2024/05/19] Update the dataset loading scripts.
- [2024/03/12] The [arxiv](https://arxiv.org/abs/2403.07721) paper is available.


<!-- ## Environment setup -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Environment setup</h2>

Run ``setup.sh`` to quickly create a conda environment that contains the packages necessary to run our scripts; activate the environment with conda activate BCI.




```
. setup.sh
```
You can also create a new conda environment and install the required dependencies by running
```
conda env create -f environment.yml
conda activate BCI

pip install wandb
pip install einops
```
Additional environments needed to run all the code:
```
pip install open_clip_torch

pip install transformers==4.28.0.dev0
pip install diffusers==0.24.0

#Below are the braindecode installation commands for the most common use cases.
pip install braindecode==0.8.1
```
<!-- ## Quick training and test  -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Quick training and test</h2>

If you want to quickly reproduce the results in the paper, please download the relevant ``preprocessed data`` and ``model weights`` from [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode) first.
#### 1.Image Retrieval
We provide the script to learn the training strategy of EEG Encoder and verify it during training. Please modify your data set path and run:
```
cd Retrieval/
python ATMS_retrieval.py --logger True --gpu cuda:0  --output_dir ./outputs/contrast
```
We also provide the script for ``joint subject training``, which aims to train all subjects jointly and test on a specific subject:
```
cd Retrieval/
python ATMS_retrieval_joint_train.py --joint_train --sub sub-01 True --logger True --gpu cuda:0  --output_dir ./outputs/contrast
```

Additionally, replicating the results of other methods (e.g. EEGNetV4) by run
```
cd Retrieval/
contrast_retrieval.py --encoder_type EEGNetv4_Encoder --epochs 30 --batch_size 1024
```

#### 2.Image Reconstruction
We provide quick training and inference scripts for ``clip pipeline`` of visual reconstruction. Please modify your data set path and run zero-shot on 200 classes test dataset:
```
# Train and generate eeg features in Subject 8
cd Generation/
python ATMS_reconstruction.py --insubject True --subjects sub-08 --logger True \
--gpu cuda:0  --output_dir ./outputs/contrast
```

```
# Reconstruct images in Subject 8
Generation_metrics_sub8.ipynb
```

We also provide scripts for image reconstruction combined ``with the low level pipeline``.
```
cd Generation/

# step 1: train vae encoder and then generate low level images
train_vae_latent_512_low_level_no_average.py

# step 2: load low level images and then reconstruct them
1x1024_reconstruct_sdxl.ipynb
```


We provide scripts for caption generation combined ``with the semantic level pipeline``.
```
cd Generation/

# step 1: train feature adapter
image_adapter.ipynb

# step 2: get caption from eeg latent
GIT_caption_batch.ipynb

# step 3: load text prompt and then reconstruct images
1x1024_reconstruct_sdxl.ipynb
```

To evaluate the quality of the reconstructed images, modify the paths of the reconstructed images and the original stimulus images in the notebook and run:
```
#compute metrics, cited from MindEye
Reconstruction_Metrics_ATM.ipynb
```

<!-- ## Data availability -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Data availability</h2>

We provide you with the ``preprocessed EEG`` and ``preprocessed MEG`` data used in our paper at [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode), as well as the raw image data.


Note that the experimental paradigms of the THINGS-EEG and THINGS-MEG datasets themselves are different, so we will provide images and data for the two datasets separately.

You can also download the relevant THINGS-EEG data set and THINGS-MEG data set at osf.io.

The raw and preprocessed EEG dataset, the training and test images are available on [osf](https://osf.io/3jk45/).
- ``Raw EEG data:`` `../project_directory/eeg_dataset/raw_data/`.
- ``Preprocessed EEG data:`` `../project_directory/eeg_dataset/preprocessed_data/`.
- ``Training and test images:`` `../project_directory/image_set/`.


The raw and preprocessed MEG dataset, the training and test images are available on [OpenNEURO](https://openneuro.org/datasets/ds004212/versions/2.0.0).





<!-- ## EEG/MEG preprocessing -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">EEG/MEG preprocessing</h2>


Modify your path and execute the following code to perform the same preprocessing on the raw data as in our experiment:
```
cd EEG-preprocessing/
python EEG-preprocessing/preprocessing.py
```

```
cd MEG-preprocessing/
MEG-preprocessing/pre_possess.ipynb
```
Also You can get the data set used in this project through the BaiduNetDisk [link](https://pan.baidu.com/s/1-1hgpoi4nereLVqE4ylE_g?pwd=nid5) to run the code.

## TODO
- [√] Release retrieval and reconstruction scripts.
- [√] Update training scripts of reconstruction pipeline.
- [ ] Adding validation sets improves performance evaluation accuracy.



<!-- ## Acknowledge -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Acknowledge</h2>

1.Thanks to Y Song et al. for their contribution in data set preprocessing and neural network structure, we refer to their work:</br>"[Decoding Natural Images from EEG for Object Recognition](https://arxiv.org/pdf/2308.13234.pdf)".</br> Yonghao Song, Bingchuan Liu, Xiang Li, Nanlin Shi, Yijun Wang, and Xiaorong Gao. 

2.We also thank the authors of [SDRecon](https://github.com/yu-takagi/StableDiffusionReconstruction) for providing the codes and the results. Some parts of the training script are based on [MindEye](https://medarc-ai.github.io/mindeye/) and [MindEye2](https://github.com/MedARC-AI/MindEyeV2). Thanks for the awesome research works.

3.Here we provide our THING-EEG dataset cited in the paper:</br>"[A large and rich EEG dataset for modeling human visual object recognition](https://www.sciencedirect.com/science/article/pii/S1053811922008758?via%3Dihub)".</br>
Alessandro T. Gifford, Kshitij Dwivedi, Gemma Roig, Radoslaw M. Cichy.


4.Another used THINGS-MEG data set provides a reference:</br>"[THINGS-data, a multimodal collection of large-scale datasets for investigating object representations in human brain and behavior.](https://elifesciences.org/articles/82580.pdf)".</br> Hebart, Martin N., Oliver Contier, Lina Teichmann, Adam H. Rockter, Charles Y. Zheng, Alexis Kidder, Anna Corriveau, Maryam Vaziri-Pashkam, and Chris I. Baker.



<!-- ## Citation -->
<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Citation</h2>

```bibtex
@inproceedings{
li2024visual,
title={Visual Decoding and Reconstruction via {EEG} Embeddings with Guided Diffusion},
author={Dongyang Li and Chen Wei and Shiying Li and Jiachen Zou and Quanying Liu},
booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
year={2024},
url={https://openreview.net/forum?id=RxkcroC8qP}
}

@article{li2024visual,
  title={Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion},
  author={Li, Dongyang and Wei, Chen and Li, Shiying and Zou, Jiachen and Liu, Quanying},
  journal={arXiv preprint arXiv:2403.07721},
  year={2024}
}
```

	a/README.md		b/README.md
1	<div align="center">	1	<div align="center">
2		2
3	<h2 style="border-bottom: 1px solid lightgray;">Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion</h2>	3	<h2 style="border-bottom: 1px solid lightgray;">Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion</h2>
4		4
5	<!-- Badges and Links Section -->	5
6	<div style="display: flex; align-items: center; justify-content: center;">
7
8	<p align="center">
9	<a href="#">
10	<p align="center">
11	<a href='https://arxiv.org/pdf/2403.07721'><img src='http://img.shields.io/badge/Paper-arxiv.2403.07721-B31B1B.svg'></a>
12	<a href='https://huggingface.co/datasets/LidongYang/EEG_Image_decode/tree/main'><img src='https://img.shields.io/badge/EEG Image decode-%F0%9F%A4%97%20Hugging%20Face-blue'></a>
13	</p>
14	</p>
15
16
17	</div>
18
19	<br/>	6	<br/>
20		7
21	</div>	8	</div>
22		9
23	<!--	10	<!--
24	<img src="bs=16_test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->	11	<img src="bs=16_test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->
25	<!--	12	<!--
26	<img src="test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->	13	<img src="test_acc.png" alt="Framework" style="max-width: 90%; height: auto;"/> -->
27		14
28	<!-- As the training epochs increases, the test set accuracy of different methods. (Top: batchsize is 16. Bottom: batchsize is 1024) -->	15	<!-- As the training epochs increases, the test set accuracy of different methods. (Top: batchsize is 16. Bottom: batchsize is 1024) -->
29		16
30	<!--	17	<!--
31	<img src="temporal_analysis.png" alt="Framework" style="max-width: 90%; height: auto;"/>	18	<img src="temporal_analysis.png" alt="Framework" style="max-width: 90%; height: auto;"/>
32	Examples of growing window image reconstruction with 5 different random seeds. -->	19	Examples of growing window image reconstruction with 5 different random seeds. -->
33		20
34		21
35	<img src="fig-framework.png" alt="Framework" style="max-width: 100%; height: auto;"/>	22	<img src="fig-framework.png" alt="Framework" style="max-width: 100%; height: auto;"/>
36	Framework of our proposed method.	23	Framework of our proposed method.
37		24
38		25
39		26
40		27
41	<!-- -->	28	<!-- -->
42	<img src="fig-genexample.png" alt="fig-genexample" style="max-width: 90%; height: auto;"/>	29	<img src="fig-genexample.png" alt="fig-genexample" style="max-width: 90%; height: auto;"/>
43		30
44	Some examples of using EEG to reconstruct stimulus images.	31	Some examples of using EEG to reconstruct stimulus images.
45		32
46		33
47	## News:	34	## News:
48	- [2024/09/26] Our paper is accepted to NeurIPS 2024.	35	- [2024/09/26] Our paper is accepted to NeurIPS 2024.
49	- [2024/09/25] We have updated the [arxiv](https://arxiv.org/abs/2403.07721) paper.	36	- [2024/09/25] We have updated the [arxiv](https://arxiv.org/abs/2403.07721) paper.
50	- [2024/08/01] Update scripts for training and inference in different tasks.	37	- [2024/08/01] Update scripts for training and inference in different tasks.
51	- [2024/05/19] Update the dataset loading scripts.	38	- [2024/05/19] Update the dataset loading scripts.
52	- [2024/03/12] The [arxiv](https://arxiv.org/abs/2403.07721) paper is available.	39	- [2024/03/12] The [arxiv](https://arxiv.org/abs/2403.07721) paper is available.
53		40
54		41
55	<!-- ## Environment setup -->	42	<!-- ## Environment setup -->
56	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Environment setup</h2>	43	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Environment setup</h2>
57		44
58	Run ``setup.sh`` to quickly create a conda environment that contains the packages necessary to run our scripts; activate the environment with conda activate BCI.	45	Run ``setup.sh`` to quickly create a conda environment that contains the packages necessary to run our scripts; activate the environment with conda activate BCI.
59		46
60		47
61		48
62		49
63	```	50	```
64	. setup.sh	51	. setup.sh
65	```	52	```
66	You can also create a new conda environment and install the required dependencies by running	53	You can also create a new conda environment and install the required dependencies by running
67	```	54	```
68	conda env create -f environment.yml	55	conda env create -f environment.yml
69	conda activate BCI	56	conda activate BCI
70		57
71	pip install wandb	58	pip install wandb
72	pip install einops	59	pip install einops
73	```	60	```
74	Additional environments needed to run all the code:	61	Additional environments needed to run all the code:
75	```	62	```
76	pip install open_clip_torch	63	pip install open_clip_torch
77		64
78	pip install transformers==4.28.0.dev0	65	pip install transformers==4.28.0.dev0
79	pip install diffusers==0.24.0	66	pip install diffusers==0.24.0
80		67
81	#Below are the braindecode installation commands for the most common use cases.	68	#Below are the braindecode installation commands for the most common use cases.
82	pip install braindecode==0.8.1	69	pip install braindecode==0.8.1
83	```	70	```
84	<!-- ## Quick training and test -->	71	<!-- ## Quick training and test -->
85	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Quick training and test</h2>	72	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Quick training and test</h2>
86		73
87	If you want to quickly reproduce the results in the paper, please download the relevant ``preprocessed data`` and ``model weights`` from [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode) first.	74	If you want to quickly reproduce the results in the paper, please download the relevant ``preprocessed data`` and ``model weights`` from [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode) first.
88	#### 1.Image Retrieval	75	#### 1.Image Retrieval
89	We provide the script to learn the training strategy of EEG Encoder and verify it during training. Please modify your data set path and run:	76	We provide the script to learn the training strategy of EEG Encoder and verify it during training. Please modify your data set path and run:
90	```	77	```
91	cd Retrieval/	78	cd Retrieval/
92	python ATMS_retrieval.py --logger True --gpu cuda:0 --output_dir ./outputs/contrast	79	python ATMS_retrieval.py --logger True --gpu cuda:0 --output_dir ./outputs/contrast
93	```	80	```
94	We also provide the script for ``joint subject training``, which aims to train all subjects jointly and test on a specific subject:	81	We also provide the script for ``joint subject training``, which aims to train all subjects jointly and test on a specific subject:
95	```	82	```
96	cd Retrieval/	83	cd Retrieval/
97	python ATMS_retrieval_joint_train.py --joint_train --sub sub-01 True --logger True --gpu cuda:0 --output_dir ./outputs/contrast	84	python ATMS_retrieval_joint_train.py --joint_train --sub sub-01 True --logger True --gpu cuda:0 --output_dir ./outputs/contrast
98	```	85	```
99		86
100	Additionally, replicating the results of other methods (e.g. EEGNetV4) by run	87	Additionally, replicating the results of other methods (e.g. EEGNetV4) by run
101	```	88	```
102	cd Retrieval/	89	cd Retrieval/
103	contrast_retrieval.py --encoder_type EEGNetv4_Encoder --epochs 30 --batch_size 1024	90	contrast_retrieval.py --encoder_type EEGNetv4_Encoder --epochs 30 --batch_size 1024
104	```	91	```
105		92
106	#### 2.Image Reconstruction	93	#### 2.Image Reconstruction
107	We provide quick training and inference scripts for ``clip pipeline`` of visual reconstruction. Please modify your data set path and run zero-shot on 200 classes test dataset:	94	We provide quick training and inference scripts for ``clip pipeline`` of visual reconstruction. Please modify your data set path and run zero-shot on 200 classes test dataset:
108	```	95	```
109	# Train and generate eeg features in Subject 8	96	# Train and generate eeg features in Subject 8
110	cd Generation/	97	cd Generation/
111	python ATMS_reconstruction.py --insubject True --subjects sub-08 --logger True \	98	python ATMS_reconstruction.py --insubject True --subjects sub-08 --logger True \
112	--gpu cuda:0 --output_dir ./outputs/contrast	99	--gpu cuda:0 --output_dir ./outputs/contrast
113	```	100	```
114		101
115	```	102	```
116	# Reconstruct images in Subject 8	103	# Reconstruct images in Subject 8
117	Generation_metrics_sub8.ipynb	104	Generation_metrics_sub8.ipynb
118	```	105	```
119		106
120	We also provide scripts for image reconstruction combined ``with the low level pipeline``.	107	We also provide scripts for image reconstruction combined ``with the low level pipeline``.
121	```	108	```
122	cd Generation/	109	cd Generation/
123		110
124	# step 1: train vae encoder and then generate low level images	111	# step 1: train vae encoder and then generate low level images
125	train_vae_latent_512_low_level_no_average.py	112	train_vae_latent_512_low_level_no_average.py
126		113
127	# step 2: load low level images and then reconstruct them	114	# step 2: load low level images and then reconstruct them
128	1x1024_reconstruct_sdxl.ipynb	115	1x1024_reconstruct_sdxl.ipynb
129	```	116	```
130		117
131		118
132	We provide scripts for caption generation combined ``with the semantic level pipeline``.	119	We provide scripts for caption generation combined ``with the semantic level pipeline``.
133	```	120	```
134	cd Generation/	121	cd Generation/
135		122
136	# step 1: train feature adapter	123	# step 1: train feature adapter
137	image_adapter.ipynb	124	image_adapter.ipynb
138		125
139	# step 2: get caption from eeg latent	126	# step 2: get caption from eeg latent
140	GIT_caption_batch.ipynb	127	GIT_caption_batch.ipynb
141		128
142	# step 3: load text prompt and then reconstruct images	129	# step 3: load text prompt and then reconstruct images
143	1x1024_reconstruct_sdxl.ipynb	130	1x1024_reconstruct_sdxl.ipynb
144	```	131	```
145		132
146	To evaluate the quality of the reconstructed images, modify the paths of the reconstructed images and the original stimulus images in the notebook and run:	133	To evaluate the quality of the reconstructed images, modify the paths of the reconstructed images and the original stimulus images in the notebook and run:
147	```	134	```
148	#compute metrics, cited from MindEye	135	#compute metrics, cited from MindEye
149	Reconstruction_Metrics_ATM.ipynb	136	Reconstruction_Metrics_ATM.ipynb
150	```	137	```
151		138
152	<!-- ## Data availability -->	139	<!-- ## Data availability -->
153	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Data availability</h2>	140	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Data availability</h2>
154		141
155	We provide you with the ``preprocessed EEG`` and ``preprocessed MEG`` data used in our paper at [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode), as well as the raw image data.	142	We provide you with the ``preprocessed EEG`` and ``preprocessed MEG`` data used in our paper at [Hugging Face](https://huggingface.co/datasets/LidongYang/EEG_Image_decode), as well as the raw image data.
156		143
157		144
158	Note that the experimental paradigms of the THINGS-EEG and THINGS-MEG datasets themselves are different, so we will provide images and data for the two datasets separately.	145	Note that the experimental paradigms of the THINGS-EEG and THINGS-MEG datasets themselves are different, so we will provide images and data for the two datasets separately.
159		146
160	You can also download the relevant THINGS-EEG data set and THINGS-MEG data set at osf.io.	147	You can also download the relevant THINGS-EEG data set and THINGS-MEG data set at osf.io.
161		148
162	The raw and preprocessed EEG dataset, the training and test images are available on [osf](https://osf.io/3jk45/).	149	The raw and preprocessed EEG dataset, the training and test images are available on [osf](https://osf.io/3jk45/).
163	- ``Raw EEG data:`` `../project_directory/eeg_dataset/raw_data/`.	150	- ``Raw EEG data:`` `../project_directory/eeg_dataset/raw_data/`.
164	- ``Preprocessed EEG data:`` `../project_directory/eeg_dataset/preprocessed_data/`.	151	- ``Preprocessed EEG data:`` `../project_directory/eeg_dataset/preprocessed_data/`.
165	- ``Training and test images:`` `../project_directory/image_set/`.	152	- ``Training and test images:`` `../project_directory/image_set/`.
166		153
167		154
168	The raw and preprocessed MEG dataset, the training and test images are available on [OpenNEURO](https://openneuro.org/datasets/ds004212/versions/2.0.0).	155	The raw and preprocessed MEG dataset, the training and test images are available on [OpenNEURO](https://openneuro.org/datasets/ds004212/versions/2.0.0).
169		156
170		157
171		158
172		159
173		160
174	<!-- ## EEG/MEG preprocessing -->	161	<!-- ## EEG/MEG preprocessing -->
175	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">EEG/MEG preprocessing</h2>	162	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">EEG/MEG preprocessing</h2>
176		163
177		164
178	Modify your path and execute the following code to perform the same preprocessing on the raw data as in our experiment:	165	Modify your path and execute the following code to perform the same preprocessing on the raw data as in our experiment:
179	```	166	```
180	cd EEG-preprocessing/	167	cd EEG-preprocessing/
181	python EEG-preprocessing/preprocessing.py	168	python EEG-preprocessing/preprocessing.py
182	```	169	```
183		170
184	```	171	```
185	cd MEG-preprocessing/	172	cd MEG-preprocessing/
186	MEG-preprocessing/pre_possess.ipynb	173	MEG-preprocessing/pre_possess.ipynb
187	```	174	```
188	Also You can get the data set used in this project through the BaiduNetDisk [link](https://pan.baidu.com/s/1-1hgpoi4nereLVqE4ylE_g?pwd=nid5) to run the code.	175	Also You can get the data set used in this project through the BaiduNetDisk [link](https://pan.baidu.com/s/1-1hgpoi4nereLVqE4ylE_g?pwd=nid5) to run the code.
189		176
190	## TODO	177	## TODO
191	- [√] Release retrieval and reconstruction scripts.	178	- [√] Release retrieval and reconstruction scripts.
192	- [√] Update training scripts of reconstruction pipeline.	179	- [√] Update training scripts of reconstruction pipeline.
193	- [ ] Adding validation sets improves performance evaluation accuracy.	180	- [ ] Adding validation sets improves performance evaluation accuracy.
194		181
195		182
196		183
197	<!-- ## Acknowledge -->	184	<!-- ## Acknowledge -->
198	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Acknowledge</h2>	185	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Acknowledge</h2>
199		186
200	1.Thanks to Y Song et al. for their contribution in data set preprocessing and neural network structure, we refer to their work:</br>"[Decoding Natural Images from EEG for Object Recognition](https://arxiv.org/pdf/2308.13234.pdf)".</br> Yonghao Song, Bingchuan Liu, Xiang Li, Nanlin Shi, Yijun Wang, and Xiaorong Gao.	187	1.Thanks to Y Song et al. for their contribution in data set preprocessing and neural network structure, we refer to their work:</br>"[Decoding Natural Images from EEG for Object Recognition](https://arxiv.org/pdf/2308.13234.pdf)".</br> Yonghao Song, Bingchuan Liu, Xiang Li, Nanlin Shi, Yijun Wang, and Xiaorong Gao.
201		188
202	2.We also thank the authors of [SDRecon](https://github.com/yu-takagi/StableDiffusionReconstruction) for providing the codes and the results. Some parts of the training script are based on [MindEye](https://medarc-ai.github.io/mindeye/) and [MindEye2](https://github.com/MedARC-AI/MindEyeV2). Thanks for the awesome research works.	189	2.We also thank the authors of [SDRecon](https://github.com/yu-takagi/StableDiffusionReconstruction) for providing the codes and the results. Some parts of the training script are based on [MindEye](https://medarc-ai.github.io/mindeye/) and [MindEye2](https://github.com/MedARC-AI/MindEyeV2). Thanks for the awesome research works.
203		190
204	3.Here we provide our THING-EEG dataset cited in the paper:</br>"[A large and rich EEG dataset for modeling human visual object recognition](https://www.sciencedirect.com/science/article/pii/S1053811922008758?via%3Dihub)".</br>	191	3.Here we provide our THING-EEG dataset cited in the paper:</br>"[A large and rich EEG dataset for modeling human visual object recognition](https://www.sciencedirect.com/science/article/pii/S1053811922008758?via%3Dihub)".</br>
205	Alessandro T. Gifford, Kshitij Dwivedi, Gemma Roig, Radoslaw M. Cichy.	192	Alessandro T. Gifford, Kshitij Dwivedi, Gemma Roig, Radoslaw M. Cichy.
206		193
207		194
208	4.Another used THINGS-MEG data set provides a reference:</br>"[THINGS-data, a multimodal collection of large-scale datasets for investigating object representations in human brain and behavior.](https://elifesciences.org/articles/82580.pdf)".</br> Hebart, Martin N., Oliver Contier, Lina Teichmann, Adam H. Rockter, Charles Y. Zheng, Alexis Kidder, Anna Corriveau, Maryam Vaziri-Pashkam, and Chris I. Baker.	195	4.Another used THINGS-MEG data set provides a reference:</br>"[THINGS-data, a multimodal collection of large-scale datasets for investigating object representations in human brain and behavior.](https://elifesciences.org/articles/82580.pdf)".</br> Hebart, Martin N., Oliver Contier, Lina Teichmann, Adam H. Rockter, Charles Y. Zheng, Alexis Kidder, Anna Corriveau, Maryam Vaziri-Pashkam, and Chris I. Baker.
209		196
210		197
211		198
212	<!-- ## Citation -->	199	<!-- ## Citation -->
213	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Citation</h2>	200	<h2 style="border-bottom: 1px solid lightgray; margin-bottom: 5px;">Citation</h2>
214		201
215	```bibtex	202	```bibtex
216	@inproceedings{	203	@inproceedings{
217	li2024visual,	204	li2024visual,
218	title={Visual Decoding and Reconstruction via {EEG} Embeddings with Guided Diffusion},	205	title={Visual Decoding and Reconstruction via {EEG} Embeddings with Guided Diffusion},
219	author={Dongyang Li and Chen Wei and Shiying Li and Jiachen Zou and Quanying Liu},	206	author={Dongyang Li and Chen Wei and Shiying Li and Jiachen Zou and Quanying Liu},
220	booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},	207	booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
221	year={2024},	208	year={2024},
222	url={https://openreview.net/forum?id=RxkcroC8qP}	209	url={https://openreview.net/forum?id=RxkcroC8qP}
223	}	210	}
224		211
225	@article{li2024visual,	212	@article{li2024visual,
226	title={Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion},	213	title={Visual Decoding and Reconstruction via EEG Embeddings with Guided Diffusion},
227	author={Li, Dongyang and Wei, Chen and Li, Shiying and Zou, Jiachen and Liu, Quanying},	214	author={Li, Dongyang and Wei, Chen and Li, Shiying and Zou, Jiachen and Liu, Quanying},
228	journal={arXiv preprint arXiv:2403.07721},	215	journal={arXiv preprint arXiv:2403.07721},
229	year={2024}	216	year={2024}
230	}	217	}
231	```	218	```