Fix metadata and improve model card (#1)
Browse files- Fix metadata and improve model card (02d82dc8726803b5fc4c42bad5957e9c8f918929)
Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>
README.md
CHANGED
|
@@ -1,5 +1,9 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: mit
|
|
|
|
|
|
|
| 3 |
tags:
|
| 4 |
- image-dehazing
|
| 5 |
- spiking-neural-network
|
|
@@ -9,15 +13,14 @@ tags:
|
|
| 9 |
- pytorch_model_hub_mixin
|
| 10 |
- model_hub_mixin
|
| 11 |
- safetensors
|
| 12 |
-
|
| 13 |
-
pipeline_tag: image-to-image
|
| 14 |
-
language:
|
| 15 |
-
- en
|
| 16 |
---
|
| 17 |
|
| 18 |
# DehazeSNN
|
| 19 |
|
| 20 |
-
> U-Net-Like Spiking Neural Networks for Single Image Dehazing
|
|
|
|
|
|
|
| 21 |
|
| 22 |
DehazeSNN integrates a U-Net-like encoder-decoder architecture with Spiking Neural Networks (SNNs), using an Orthogonal Leaky-Integrate-and-Fire Block (OLIFBlock) for efficient cross-channel communication. This yields competitive dehazing quality with fewer parameters and MACs compared to Transformer-based methods.
|
| 23 |
|
|
@@ -40,7 +43,7 @@ DehazeSNN integrates a U-Net-like encoder-decoder architecture with Spiking Neur
|
|
| 40 |
|
| 41 |
```bash
|
| 42 |
# Clone the DehazeSNN repository (for model code + custom CUDA kernels)
|
| 43 |
-
git clone https://github.com/
|
| 44 |
cd DehazeSNN
|
| 45 |
|
| 46 |
# Create environment (requires CUDA 12.x)
|
|
@@ -90,10 +93,10 @@ Image.fromarray(output).save("dehazed_image.jpg")
|
|
| 90 |
|
| 91 |
## Requirements
|
| 92 |
|
| 93 |
-
- **CUDA GPU required**: The custom LIF CUDA kernels require an NVIDIA GPU with CUDA support
|
| 94 |
-
- **CuPy**: `cupy-cuda12x` (for CUDA 12.x) - CPU-only inference is **not supported
|
| 95 |
-
- PyTorch >= 2.1 with CUDA 12.1
|
| 96 |
-
- Python 3.11 recommended
|
| 97 |
|
| 98 |
## Model Sizes
|
| 99 |
|
|
@@ -119,10 +122,4 @@ If you find this work useful, please cite our paper:
|
|
| 119 |
|
| 120 |
## License
|
| 121 |
|
| 122 |
-
This project is released under the MIT License.
|
| 123 |
-
|
| 124 |
-
## Links
|
| 125 |
-
|
| 126 |
-
- [GitHub Repository](https://github.com/FengShaner/DehazeSNN)
|
| 127 |
-
- [Paper (IEEE IJCNN 2025)](https://doi.org/10.1109/IJCNN64981.2025.11228727)
|
| 128 |
-
- [Pretrained Models & Results (Zenodo)](https://doi.org/10.5281/zenodo.15486831)
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
license: mit
|
| 5 |
+
pipeline_tag: image-to-image
|
| 6 |
+
library_name: pytorch
|
| 7 |
tags:
|
| 8 |
- image-dehazing
|
| 9 |
- spiking-neural-network
|
|
|
|
| 13 |
- pytorch_model_hub_mixin
|
| 14 |
- model_hub_mixin
|
| 15 |
- safetensors
|
| 16 |
+
arxiv: 2512.23950
|
|
|
|
|
|
|
|
|
|
| 17 |
---
|
| 18 |
|
| 19 |
# DehazeSNN
|
| 20 |
|
| 21 |
+
> **U-Net-Like Spiking Neural Networks for Single Image Dehazing**
|
| 22 |
+
|
| 23 |
+
[[📄 Paper](https://huggingface.co/papers/2512.23950)] [[💻 GitHub](https://github.com/HaoranLiu507/DehazeSNN)]
|
| 24 |
|
| 25 |
DehazeSNN integrates a U-Net-like encoder-decoder architecture with Spiking Neural Networks (SNNs), using an Orthogonal Leaky-Integrate-and-Fire Block (OLIFBlock) for efficient cross-channel communication. This yields competitive dehazing quality with fewer parameters and MACs compared to Transformer-based methods.
|
| 26 |
|
|
|
|
| 43 |
|
| 44 |
```bash
|
| 45 |
# Clone the DehazeSNN repository (for model code + custom CUDA kernels)
|
| 46 |
+
git clone https://github.com/HaoranLiu507/DehazeSNN.git
|
| 47 |
cd DehazeSNN
|
| 48 |
|
| 49 |
# Create environment (requires CUDA 12.x)
|
|
|
|
| 93 |
|
| 94 |
## Requirements
|
| 95 |
|
| 96 |
+
- **CUDA GPU required**: The custom LIF CUDA kernels require an NVIDIA GPU with CUDA support.
|
| 97 |
+
- **CuPy**: `cupy-cuda12x` (for CUDA 12.x) - CPU-only inference is **not supported**.
|
| 98 |
+
- PyTorch >= 2.1 with CUDA 12.1.
|
| 99 |
+
- Python 3.11 recommended.
|
| 100 |
|
| 101 |
## Model Sizes
|
| 102 |
|
|
|
|
| 122 |
|
| 123 |
## License
|
| 124 |
|
| 125 |
+
This project is released under the MIT License.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|