Enhance dataset card: Add paper/code links, tasks, sample usage, correct license, and citations

README.md

@@ -1,29 +1,154 @@
 ---
-license: apache-2.0
+license: mit
+task_categories:
+- image-feature-extraction
+- image-to-image
+- image-to-text
+language:
+- en
+tags:
+- neuroscience
+- brain-decoding
+- eeg
+- meg
+- fmri
+- multimodal
+- bci
 ---
 
+# BrainFLORA: Uncovering Brain Concept Representation via Multimodal Neural Embeddings
+
+This repository contains the images and preprocessed data from THINGS-EEG1, THINGS-fMRI, THINGS-MEG dataset, associated with the paper [BrainFLORA: Uncovering Brain Concept Representation via Multimodal Neural Embeddings](https://huggingface.co/papers/2507.09747).
+
+*   **Paper**: [https://huggingface.co/papers/2507.09747](https://huggingface.co/papers/2507.09747)
+*   **Code**: [https://github.com/ncclab-sustech/BrainFLORA](https://github.com/ncclab-sustech/BrainFLORA)
+
 ## 📖 Dataset Description
 
-This dataset contains the images and preprocessed data from THINGS-EEG1，THINGS-fMRI，THINGS-MEG dataset. 
+This dataset contains the images and preprocessed data from THINGS-EEG1, THINGS-fMRI, THINGS-MEG dataset, which can be directly used for training.
+
+To download the raw data, you can follow these links:
 
+*   **THINGS-EEG1**: [Download](https://openneuro.org/datasets/ds003825/versions/1.1.0)
+*   **THINGS-EEG2**: [Download](https://osf.io/3jk45/)
+*   **THINGS-MEG**: [Download](https://openneuro.org/datasets/ds004212/versions/2.0.0)
+*   **THINGS-fMRI**: [Download](https://openneuro.org/datasets/ds004192/versions/1.0.7)
+*   **THINGS-Images**: [Download](https://osf.io/rdxy2)
 
 ## 🧾 Supported Tasks
 
-- **Task 1** (e.g., THINGS-EEG1)
-  - **Input**: EEG
-  - **Output**: Image label
-  - **Evaluation metric**: Accuracy
+-   **Visual Retrieval**
+    -   **Input**: EEG, MEG, or fMRI brain signals
+    -   **Output**: Image label (retrieved image)
+    -   **Evaluation metric**: Accuracy
 
-- **Task 2** (e.g., THINGS-MEG)
-  - **Input**: MEG
-  - **Output**: Image label
-  - **Evaluation metric**: Accuracy
+-   **Visual Reconstruction**
+    -   **Input**: EEG, MEG, or fMRI brain signals
+    -   **Output**: Reconstructed image
+    -   **Evaluation metric**: Not specified in current card, but typically image similarity metrics.
 
-- **Task 3** (e.g., THINGS-fMRI)
-  - **Input**: fMRI
-  - **Output**: Image label
-  - **Evaluation metric**: Accuracy
+-   **Visual Captioning**
+    -   **Input**: EEG, MEG, or fMRI brain signals
+    -   **Output**: Text caption describing the image
+    -   **Evaluation metric**: Not specified in current card, but typically text generation metrics like BLEU, ROUGE, etc.
 
 ---
 
+## 🚀 Sample Usage
+
+The following snippets demonstrate how to use the BrainFLORA code for training and evaluation of different tasks. First, ensure your environment is set up by following the instructions in the [GitHub repository](https://github.com/ncclab-sustech/BrainFLORA#%EF%B8%8Fenvironment-setup).
+
+### Environment Setup
+
+Run `setup.sh` to quickly create a conda environment that contains the packages necessary to run our scripts; activate the environment with `conda activate BrainFLORA`.
+
+```bash
+. setup.sh
+```
+
+You can also create a new conda environment and install the required dependencies by running:
+
+```bash
+conda env create -f environment.yml
+conda activate BrainFLORA
+```
+
+### Visual Retrieval
+
+Train modality encoders for joint subject training on the THINGS-EEG2 dataset (modify your dataset path):
+
+```python
+cd Retrieval/
+python retrieval_joint_train_medformer.py --logger True --gpu cuda:0  --output_dir ./outputs/contrast
+```
+
+Replicate results for other modalities (e.g., MEG, fMRI):
+
+```python
+cd Retrieval/
+python retrieval_joint_train_MEG_rerank_medformer.py --logger True --gpu cuda:0  --output_dir ./outputs/contrast
+```
+
+Evaluate the models using the provided notebook:
+
+```bash
+cd eval/
+# Open FLORA_inference.ipynb
+```
+
+### Visual Reconstruction
+
+Train and get multimodal neural embeddings aligned with CLIP embeddings (for high-level and low-level visual reconstruction pipeline):
+
+```python
+python train_unified_encoder_highlevel_diffprior.py --modalities ['eeg', 'meg', 'fmri'] --gpu cuda:0  --output_dir ./outputs/contrast
+```
+
+Reconstruct images by assigning modalities and subjects:
+
+```python
+python FLORA_inference_reconst.py
+```
+
+### Visual Captioning
+
+**Step 1**: Train feature adapter (same as the first step in Visual Reconstruction):
+
+```python
+python train_unified_encoder_highlevel_diffprior.py --modalities ['eeg', 'meg', 'fmri'] --gpu cuda:0  --output_dir ./outputs/contrast
+```
+
+**Step 2**: Get captions from the prior latent space:
+
+```bash
+# Open FLORA_inference_caption.ipynb
+```
+
+## 📚 Citations
+
+If you find our work useful, please consider citing:
+
+```bibtex
+@article{li2025brainflora,
+  title={BrainFLORA: Uncovering Brain Concept Representation via Multimodal Neural Embeddings},
+  author={Li, Dongyang and Qin, Haoyang and Wu, Mingyang and Wei, Chen and Liu, Quanying},
+    journal={arXiv preprint arXiv:2507.09747},
+  year={2025}
+}
 
+@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={Advances in Neural Information Processing Systems},
+  volume={37},
+  pages={102822--102864},
+  year={2024}
+}
+@inproceedings{wei2024cocog,
+  title={CoCoG: controllable visual stimuli generation based on human concep08/03/2024t representations},
+  author={Wei, Chen and Zou, Jiachen and Heinke, Dietmar and Liu, Quanying},
+  booktitle={Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence},
+  pages={3178--3186},
+  year={2024}
+}
+```