Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/byebyebruce/florence2-app
Microsoft Florence-2(variety of vision Tasks) demo.
https://github.com/byebyebruce/florence2-app
Last synced: about 5 hours ago
JSON representation
Microsoft Florence-2(variety of vision Tasks) demo.
- Host: GitHub
- URL: https://github.com/byebyebruce/florence2-app
- Owner: byebyebruce
- Created: 2024-06-27T04:00:07.000Z (5 months ago)
- Default Branch: master
- Last Pushed: 2024-06-27T04:13:23.000Z (5 months ago)
- Last Synced: 2024-06-27T05:50:49.577Z (5 months ago)
- Language: Python
- Homepage:
- Size: 41 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Florence2 App
> 本项目是将[Microsoft Florence2](https://huggingface.co/microsoft/Florence-2-large)([Paper](https://arxiv.org/abs/2311.06242) | [Model](https://huggingface.co/microsoft/Florence-2-large) | [Notebook](https://huggingface.co/microsoft/Florence-2-large/blob/main/sample_inference.ipynb)) 模型封装成 Flask 的 API 和 CLI 工具 。它可以完成多种[视觉任务](#支持视觉任务列表)，如生成描述，执行目标检测、OCR 等。
## 功能特点
- **API 模式**：提供一个API Server来处理图像任务请求。
- **CLI 模式**：用户通过命令行直接输入任务和图像路径。
## 安装
1. **克隆仓库**：
```sh
git clone https://github.com/byebyebruce/florence2-app.git
cd florence2-app
```
2. **创建虚拟环境**：
```sh
conda create -n florence2-app python=3.11
conda activate florence2-app
```
3. **安装依赖**：
```sh
pip install -r requirements.txt
```
## 使用方法
### 运行 API Server
```sh
python3 main.py api
```
#### 请求示例：
1. 目标检测
```sh
curl -X POST http://localhost:5000/api/predict \
-F "image=@testdata/car.jpg" \
-F "task=OD"
```
2. 图片描述
```sh
curl -X POST http://localhost:5000/api/predict \
-F "image=@testdata/car.jpg" \
-F "task=CAPTION"
```
3. 特定区域分割
```sh
curl -X POST http://localhost:5000/api/predict \
-F "image=@testdata/car.jpg" \
-F "task=REFERRING_EXPRESSION_SEGMENTATION" \
-F "prompt=the car"
```
### 运行 CLI 工具
```sh
python3 main.py cli OD ./testdata/car.jpg
```
## 支持视觉任务列表
| Task Prompt | Description |
|--------------------------------------|---------------------------------------|
| DETAILED_CAPTION | 详细图像描述 |
| MORE_DETAILED_CAPTION | 更详细图像描述 |
| CAPTION_TO_PHRASE_GROUNDING | 图像描述到定语 |
| DENSE_REGION_CAPTION | 密集区域描述 |
| OD | 目标检测 |
| REGION_PROPOSAL | 候选区域 |
| OCR | 字符识别 |
| OCR_WITH_REGION | 区域字符识别 |
| REFERRING_EXPRESSION_SEGMENTATION | REFERRING_EXPRESSION_SEGMENTATION |
| REGION_TO_SEGMENTATION | 区域分割 |
| OPEN_VOCABULARY_DETECTION | 开放词汇下的目标检测 |
| REGION_TO_CATEGORY | 区域类别 |
| REGION_TO_DESCRIPTION | 区域描述 |