mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-18 22:44:39 +08:00
English | 简体中文
BlazeFace Ready-to-deploy Model
- BlazeFace deployment model implementation comes from BlazeFace,and Pre-training model based on WiderFace
- （1）Provided in Official library *.params, could deploy after operation [export_model.py](#Export PADDLE model);
- （2）Developers can train BlazeFace model based on their own data according to export_model. pyAfter exporting the model, complete the deployment。
Export PADDLE model
Visit BlazeFace Github library, download and install according to the instructions, download the . yml and . params model parameters, and useexport_ Model. pygets thepadmodel file. yml,. pdiparams,. pdmodel .
- Download BlazeFace model parameter file
| Network structure | input size | number of pictures/GPU | learning rate strategy | Easy/Media/Hard Set | prediction delay (SD855) | model size (MB) | download | configuration file |
|---|---|---|---|---|---|---|---|---|
| BlazeFace | 640 | 8 | 1000e | 0.885 / 0.855 / 0.731 | - | 0.472 | Download link | Config file |
| BlazeFace-FPN-SSH | 640 | 8 | 1000e | 0.907 / 0.883 / 0.793 | - | 0.479 | Download link | Config file |
- Export paddle-format file
python tools/export_model.py -c configs/face_detection/blazeface_1000e.yml -o weights=blazeface_1000e.pdparams --export_serving_model=True
Detailed Deployment Tutorials
Release Note
- This tutorial and related code are written based on BlazeFace