AMB82 AI Model Conversion

Note

Please Log in to access the page.

Refer to the following table for online conversion supported models currently.
ModelsBasic functionsRequires files
yolov3-tiny, darknetObject Detection“.cfg”, “.weights”
yolov4-tiny, darknetObject Detection“.cfg”, “.weights”
yolov7-tiny, darknetObject Detection“.cfg”, “.weights”
yolov7-tiny, pytorchObject Detection“.pt”
scrfd/mobilefacenetFace Detection & Recognition“.pt” or “.onnx”
yamnetSound Classification“.h5”
CNN Gray/RGBImage Classification“.h5” or “.onnx”
The quantize images are optional for uploading that up to 10.

For CNN Gray/RGB, please refer to the following instructions to get correct converted models.

  • When training model: supported tensorflow version up to 2.14.1

get-start-1

  • When saving model: set flag as “include_optimizer=False”

get-start-1

After AI Model Conversion, there will be download link send out by email.
For more information, please refer to the Customized AI model Installation Guide at https://www.amebaiot.com/en/ameba-arduino-summary/.
Please refer to AMB82 MobileFaceNet Convert To ONNX at https://www.amebaiot.com/en/amebapro2-mobilefacenet-convert-to-onnx/.
Please refer to AMB82 SCRFD Convert To ONNX at https://www.amebaiot.com/en/amebapro2-scrfd-convert-to-onnx/.
Please refer to AMB82 Yolov7-Tiny for Pytorch version, the official site at https://github.com/WongKinYiu/yolov7.
Please confirm that QQ communication software is installed