Skip to content

using ncnn, on android, detection, model size only 11MB, 10fps on 845/cpu

Notifications You must be signed in to change notification settings

JuZiSYJ/MobilenetSSD_Android

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MobilenetSSD_Android

using ncnn, on android,detection,surport arm v7/v8/x86, surport vulkan on gpu

environment

ncnn
AndroidStudio3.2(NDK19)
Qualcomm845

model

using pretrained MobilenetSSD on VOC,bychuanqi305,mAP=0.72
notice:the model in ncnn/example/mobilenetssd is too old to perform,please use ncnn-assert
pipeline: model on PC --> ONNX ---> model on ncnn(.bin,,param) --> inference on Android
converted model on ncnn is from ncnn-assert

organization

using ncnn and opencv on android
the sim-overall android-organization is here

organization

result

infer time on 845 in float32 is aroud 100ms, depending on the object
845-cpu 845-gpu

ncnn is good in optimazing on float32

apk

the apk is available on BAaiduYun,password is 3alh

reference

(https://github.com/nihui/ncnn-android-squeezenet)
(https://blog.csdn.net/qq_33431368/article/details/84990390)

About

using ncnn, on android, detection, model size only 11MB, 10fps on 845/cpu

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published