重现步骤
I want to run retina face detction model in k230 board using canmv ide.
As per suggested in https://developer.canaan-creative.com/ai_docs/en/main/Introduction_to_This_Document.html i am trying to convert onnx model to kmodel; that conversion is successful and cosine similarities is as 0.97337914,0.9999999,0.96764433.
期待结果和实际结果
So using this converted kmodel i should able to run kmodel in canmv ide but it is not working.
Model which is already present in image(face_detection_640) is working although, but as given in AI development tutorial https://github.com/JayL323/K230_AI_Demo_Development_Process_Analysis/blob/main/kmodel_related/kmodel_export/face_detection/onnx/k230_face_detection_data_100_640.kmodel also not works.
软硬件版本信息
Hardware Board : CanMV-K230-V1.1
Software Img: CanMV-K230_V1P0_P1_micropython_v1.2.2-0-g4b8cae1_nncase_v2.9.0.img
Windows version : Microsoft windows 10 pro
Python Version : 3.10.11
NNcase version : 2.9.0
错误日志
Device get stuck after first camera frame.
尝试解决过程
Micropython code is face detection example code, i have changes anchore file location, kmodel location, input size and anchore ength.
Model conversion code is as per suggested in documents.
补充材料
Please help me to run converted kmodel from onnx in canmv ide.
