Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
huolongshe c18ee56454 | 2 months ago | |
---|---|---|
app | 9 months ago | |
demo_data | 9 months ago | |
docs | 9 months ago | |
.gitignore | 9 months ago | |
Dockerfile | 9 months ago | |
LICENSE | 9 months ago | |
README.md | 2 months ago | |
application.yml | 9 months ago | |
build-docker.sh | 9 months ago | |
pack_model.py | 9 months ago | |
pip-install-reqs.sh | 9 months ago | |
requirements.txt | 9 months ago | |
run_model_server.py | 9 months ago |
针对人脸识别系统中经常遇到的低质量、噪声、甚至不同数据分布(out of distribution, OOD)的数据带来的问题,用基于概率的视角分析损失函数中温度调节参数和分类不确定度的内在关系,同时该不确定度服从一个先验分布。从而可以稳定训练,以及在部署时提供一个对不确定度的度量分值,帮助建立更鲁棒的人脸识别系统。 主要贡献点如下: (1)基于概率视角,揭示了损失函数中温度调节参数和分类不确定度的内在关系,通过提出的Random Temperature Scaling (RTS) 来训练更可靠的人脸识别模型。 (2)在训练阶段,RTS可以调节干净数据和噪声数据对训练的影响以得到更稳定的训练过程和更好的识别效果。 (3)在测试阶段,RTS可以提供一个不需要通过额外数据训练的不确定度分值,来分辨出不确定的、低质量的以及不同数据分布(out of distribution, OOD)的样本,以建立更鲁棒的人脸识别系统。
本模型基于 ServiceBoot微服务引擎 进行服务化封装,参见: 《CubeAI模型开发指南》
$ sh pip-install-reqs.sh
$ serviceboot start
或
$ python3 run_model_server.py
一键式本地容器化部署和运行,参见: 《CubeAI模型独立部署指南》 或 CubeAI Docker Builder
本模型服务可一键发布至 CubeAI智立方平台 进行共享和部署,参见: 《CubeAI模型发布指南》
本模型提供了4个API接口:
API接口1:
API端点: /api/data
HTTP方法: POST
HTTP请求体:
{
"action": "add_face"
"args": {
"name": <姓名>,
"img": <压缩图像的base64编码字符串(或其Data URL表示)>
}
}
HTTP响应体:
{
"status": "ok"|"err",
"value": 1(添加人脸成功)|0(图像中无人脸或有多余一个人脸)|-1(已经存在相似人脸)
}
API接口2:
API端点: /api/data
HTTP方法: POST
HTTP请求体:
{
"action": "predict"
"args": {
"img": <压缩图像的base64编码字符串(或其Data URL表示)>
}
}
HTTP响应体:
{
"status": "ok"|"err",
"value": [<识别结果>, <gps信息>, <带姓名标注的base64编码图像URL>]
}
API接口3:
API端点: /api/stream/predict
HTTP方法: POST
HTTP请求体:
<二进制编码的压缩图像字节流>
HTTP响应体:
{
"status": "ok"|"err",
"value": [<识别结果>, <gps信息>, <带姓名标注的base64编码图像URL>]
}
API接口4:
API端点: /api/file/predict
HTTP方法: POST
HTTP请求体:
<用于HTTP文件上传的XHR格式请求体>
HTTP响应体:
{
"status": "ok"|"err",
"value": [<识别结果>, <gps信息>, <带姓名标注的base64编码图像URL>]
}
No Description
Python Shell Dockerfile Text
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》