请问工程中提供的yolo3_frame_test_public_maixpy是如何往MAIXPY中部署并调用的？MAIXPY支持c语言程序吗？

[sunrong0511]

No description provided.

[sunrong0511]

另，请问您说明文档中提到的My Demo use yolo_mobilev1 0.75，这个模型训练完成后的.h5文件大小是多少？.tflite文件大小是多少？裁剪后的kmodel大小是多少？

[sunrong0511]

@zhen8838 在吗？求教

[zhen8838]

h5 大概13mb，kmodel大概3.7mb。这里的maixpy demo是直接编译c代码烧录进去的。

[sunrong0511]

h5 大概13mb，kmodel大概3.7mb。这里的maixpy demo是直接编译c代码烧录进去的。

是烧录到maixgo里吗？

[zhen8838]

是的，用kflash烧录

[sunrong0511]

是的，用kflash烧录

没有调用相机采图，只是对预存的image进行预测并显示，对么？

[zhen8838]

嗯

[sunrong0511]

嗯

是因为demo中的采图代码不支持maixpy的dvp吗？

[zhen8838]

maixpy的dvp我没有去试过,你可以参考他们官方的dvp代码.应该差别不大.

[easy-and-simple]

@sunrong0511
Guy that is replaying here have no any idea what is this maixpy, so he is talking only nonsenses
Ofcourse maixpy can't run c code, but even if it was able to run C code, models produced by this repo are too big for maixpy and can't be loaded in memory and also this repo creates yolo v3 modelas while maixpy supports only yolo v2 models
This repo can't produce 1 output models, and there is no documentation how to run models that have more than 1 output on maixpy

[sunrong0511]

@easy-and-simple I have tested and verified that “\yolo3_frame_test_public_maixpy\” can be arrange and run on maixgo. but there are indeed some problems such as outputs number and result. I think @zhen8838 just make some simple modifies on the base of yolov2 in “\yolo3_frame_test_public_maixpy\” ,but not really yolov3, am I right?

[zhen8838]

模型是yolov3的,因为他是多个输出层,训练方式也是yolov3的. 只是我在这里放的后处理代码是官方yolov2改过去的,无法实时运行. gif里面实时运行的后处理代码是我重新写的,没有开源.

[sunrong0511]

模型是yolov3的,因为他是多个输出层,训练方式也是yolov3的. 只是我在这里放的后处理代码是官方yolov2改过去的,无法实时运行. gif里面实时运行的后处理代码是我重新写的,没有开源.

Well，I guess it right. but thank you very much all the time，have learn lots from your repo.

[uniartisan]

@sunrong0511
Guy that is replaying here have no any idea what is this maixpy, so he is talking only nonsenses
Ofcourse maixpy can't run c code, but even if it was able to run C code, models produced by this repo are too big for maixpy and can't be loaded in memory and also this repo creates yolo v3 modelas while maixpy supports only yolo v2 models
This repo can't produce 1 output models, and there is no documentation how to run models that have more than 1 output on maixpy

I have met the same issue. It seems that this repo can not be deployed in MaixPy. MaixPY can only run yolo v2 models now.
Have you realized running yolo in maixpy?