EMS-YOLO - GitLite

KO | EN

gitlite — search

#javascript #python #go #golang #android #machine-learning #linux #macos #nodejs #windows #awesome #vue #pytorch #security

EMS-YOLO

(★ 196)

Offical implementation of "Deep Directly-Trained Spiking Neural Networks for Object Detection" (ICCV2023)

파일 탐색기

최종 버전 다운로드 (.zip)

- - hyp.scratch.yaml
- - get_coco.sh
- coco.yaml
- - val.cpython-310.pyc
  - val.cpython-39.pyc
- - coco.yaml
  - gen1.yaml
- - __init__.py
  - common.py
  - experimental.py
  - res10-ee.yaml
  - res18-ee.yaml
  - res18-eebk.yaml
  - res18-sew.yaml
  - resnet10.yaml
  - resnet18.yaml
  - resnet34-cat.yaml
  - resnet34.yaml
  - tf.py
  - yolo.py
- - __init__.py
  - activations.py
  - augmentations.py
  - autoanchor.py
  - autobatch.py
  - callbacks.py
  - datasets.py
  - datasets_g1T.py
  - downloads.py
  - general.py
  - give_g1_data.py
  - loss.py
  - metrics.py
  - plots.py
  - torch_utils.py
- calculate_fr.py
- detect.py
- get_gen1_data.py
- train_g1.py
- val.py
- __init__.py
- common.py
- experimental.py
- res10-ee.yaml
- res18-ee.yaml
- res18-eebk.yaml
- res18-sew.yaml
- resnet10.yaml
- resnet18.yaml
- resnet34-cat.yaml
- resnet34.yaml
- tf.py
- yolo.py
- - hyp.yaml
  - opt.yaml
  - results.png
- - __init__.py
- __init__.py
- activations.py
- augmentations.py
- autoanchor.py
- autobatch.py
- callbacks.py
- datasets.py
- downloads.py
- general.py
- loss.py
- metrics.py
- plots.py
- torch_utils.py
detect.py
environment.yml
export.py
hubconf.py
LICENSE
README.md
requirements.txt
train.py
val.py
visi_img.py

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<div align="center">

## <div align="center">Deep Directly-Trained Spiking Neural Networks for Object Detection [(ICCV2023)](https://openaccess.thecvf.com/content/ICCV2023/html/Su_Deep_Directly-Trained_Spiking_Neural_Networks_for_Object_Detection_ICCV_2023_paper.html)</div>
</div>

### Requirements

The code has been tested with pytorch=1.10.1,py=3.8, cuda=11.3, cudnn=8.2.0_0 . The conda environment can be copied directly via <b>environment.yml</b>. Some additional dependencies can be found in the  <b>environment.txt</b>.

<details open>
<summary>Install</summary>

```bash
$ git clone https://github.com/BICLab/EMS-YOLO.git
$ pip install -r requirements.txt
```

</details>

### Pretrained Checkpoints

We provide the best and the last trained model based on EMS-Res34 on the COCO dataset.

`detect.py` runs inference on a variety of sources, downloading models automatically from
the [COCO_EMS-ResNet34](https://drive.google.com/drive/folders/1mry8sdED6ncqxajmQROKBECpcrmXStpB?usp=sharing) .

The relevant parameter files are in the `runs/train`.

### Training & Addition
<details open>
<summary>Train</summary>

The relevant code for the Gen1 dataset is at `/g1-resnet`. It needs to be replaced or added to the appropriate root folder.

For gen1 dataset:

```python
python path/to/train_g1.py --weights ***.pt --img 640
```
For coco dataset:
```python
python train.py
```
</details>

Calculating the spiking rate:

Dependencies can be downloaded from [Visualizer](https://github.com/luo3300612/Visualizer).
```python
python calculate_fr.py
```

### Supplementary Note1
When running on the COCO dataset, make sure to modify line 118 in the yolo.py file to:

m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))])  # forward

### Supplementary Note2
The parameter size of ResNet34 was incorrectly reported in the paper; the correct value should be over 30M.

### Contact Information

```shell
@inproceedings{su2023deep,
  title={Deep Directly-Trained Spiking Neural Networks for Object Detection},
  author={Su, Qiaoyi and Chou, Yuhong and Hu, Yifan and Li, Jianing and Mei, Shijie and Zhang, Ziyang and Li, Guoqi},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={6555--6565},
  year={2023}
}
```

<p>
YOLOv3  is a family of object detection architectures and models pretrained on the COCO dataset, and represents <a href="https://ultralytics.com">Ultralytics</a>
 open-source research into future vision AI methods, incorporating lessons learned and best practices evolved over thousands of hours of research and development. 
 
 <b>Our code is also implemented in this framework, so please remember to cite their work.</b>
</p>