CoolPI-AI is a free and open source repository containing a series of AI demos. Developers can quickly deploy AI models on CoolPI with reference to demo to improve work efficiency.
In CoolPI-AI, we provide deployment examples of the following models. We hope this will help you better use CoolPI.
| Mission Scenario | Model | Coolpi CPU / NPU speed(ms) |
|---|---|---|
| Classification | ResNet50_vd | -/33 |
| Detection | Picodet | 162/112 |
| Detection | YOLOV5-s-relu | -/57 |
| Detection | YOLOV8-n | 192/119 |
| Character Segmentation | PPHumanSegV2Lite-Portrait | 133/43 |
| Face Detection | SCRFD-2.5G-kps-640 | 108/42 |
| Face FaceRecognition | AdaFace/ArcFace/CosFace(R18) | 81/12 |
Before using CoolPI-AI, you need to build a development environment. Please enter the following command in the root directory of Repo:
source third_party/fastdeploy-develop/fastdeploy_init.sh
sudo cp third_party/fastdeploy-develop/fastdeploy_libs.conf /etc/ld.so.conf.d/
sudo ldconifg