File size: 3,572 Bytes
f36bbc5 0af0696 0eabbc1 0af0696 74a523c 0af0696 74a523c d97825c 74a523c 19ee8c3 74a523c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 |
---
title: SpecDETR Gradio App
emoji: 🚀
colorFrom: indigo
colorTo: blue
sdk: gradio
sdk_version: "3.50.2"
app_file: app.py
pinned: false
---
Check out the configuration reference at https://huggingface.co./docs/hub/spaces-config-reference
# SpecDETR Hugging Face Space
This repository contains a Gradio demo for SpecDETR, supporting both image and video detection.
## Usage
- Upload your model config (.py) and checkpoint (.pth) files to the interface.
- Test detection on images or videos using the Gradio UI.
## Files
- `app.py`: Gradio app entry point.
- `requirements.txt`: Python dependencies for Hugging Face Spaces.
- `download_spod.py`: (Optional) Script for dataset download and config update.
## Deployment
1. Push this repo to your Hugging Face Space.
2. The Space will auto-build and launch the Gradio demo.
<<<<<<< HEAD
## Note
- If you need to add default configs or checkpoints, place them in the repo and update `app.py` accordingly.
=======
The code is built upon mmdetection 3.0.0 and runs on the Ubuntu system.
1. Create a new conda environment and activate the environment. Requires Python>=3.7.
2. Install Pytorch. Requires torch>=1.8.
3. Install mmengine and mmcv.
```
pip install mmengine==0.7.3
pip install mmcv>=2.0.0
```
4. Clone this repository:
```
SpecDETR_ROOT=/path/to/clone/SpecDETR
git clone https://github.com/ZhaoxuLi123/SpecDETR $SpecDETR_ROOT
```
5. Compile and install mmdet. If mmdet>=3.0.0 is already installed in the environment, skip this step.
```
cd SpecDETR_ROOT
pip install -v -e .
```
## SPOD Dataset Preparation
1. Download SPOD_30b_8c.zip from [Baidu Netdisk](https://pan.baidu.com/s/1vw23KWPSus2Yuj-CA1URnw?pwd=1234) (code: 1234) or [Google Drive](https://drive.google.com/file/d/1wfoLkfZOxxEtuDyDSWCnAq-N8HZVKdQ1/view?usp=drive_link).
2. Unzip SPOD_30b_8c.zip to the specified path.
```
SPOD_ROOT=/path/to/unzip/SPOD_dataset
unzip -d SPOD_ROOT SPOD_30b_8c.zip
```
3. Update the dataset configuration file ./configs/_base_/datasets/hsi_detection.py
```
Line3 data_root = 'SPOD_ROOT/SPOD_30b_8c/'
```
## Benchmark Evaluation and Training
### Pre-trained Weights Preparation
1. Download the pre-trained weights file SpecDETR_100e.pth from [Baidu Netdisk](https://pan.baidu.com/s/12-33-sCQWcMYUy5QU7rVNQ?pwd=1234) (code: 1234) or [Google Drive](https://drive.google.com/file/d/1h6_MzTb_jQ-7I09x2qfg4vci4hA50hhY/view?usp=drive_link).
2. Place the file SpecDETR_100e.pth under ./work_dirs/SpecDETR/
### Evaluation
1. Obtain detection results and evaluate AP.
```
python test.py
```
2. Evaluate inference speed FPS.
```
python train.py
```
3. Evaluate flops.
```
python benchmark.py
```
4. Retrain the network from scratch.
```
python train.py
```
Note: Although we have fixed all random seeds, there may still be slight differences in AP performance after retraining from scratch each time. This difference originates from the underlying mechanism of CUDA.
## Citation
If the work or the code is helpful, please cite the paper:
```
@article{li2024specdetr,
title={SpecDETR: A Transformer-based Hyperspectral Point Object Detection Network},
author={Li, Zhaoxu and An, Wei and Guo, Gaowei and Wang, Longguang and Wang, Yingqian and Lin, Zaiping},
journal={arXiv preprint arXiv:2405.10148},
year={2024}
}
```
## Contact
For further questions or details, please directly reach out to [email protected].
>>>>>>> dc3c711 (Update README.md)
|