Well done! This is very clear and to the point, even with typos that happen to us all. Lol 😂 If I could give you two thumbs up I would. 👍👍 +1 Subscribed!
Excellent tutorial! Thank you so much! I have a question, Would it be possible to detect motion and speed by processing camera input using OpenCv? Does it have the functionality for that? As an example: your device is capturing the environment while it is inside a moving vehicle. Would it be possible to image process the environtment and determine the vehicle is moving and also gauge its speed? Thank you!
If you can find an algorithm to gauge car speed from a stream of images then doing this with opencv on Android should be possible, I'm not familiar with such an algorithm... If you want to see how to process camera frames in opencv you can follow my playlist "crossolatform tensorflow lite" I hope to upload soon a video about object detection from the camera stream, which has some part using opencv.
@@TheCodingNotebook That would be great! Yeah any info about how to process camera frames would help. Thank you for the answer. There are a couple of algorithms that can extrapolate speed using parallax. I am hoping OpenCv can provide the pixel tracking that is needed for those algorithms.
Thank you so much for this video, in depth explanation indeed. Just asking would this image processing is better than image processing using java/kotlin in term of performance?
I never tried to do any image processing with java/kotlin, reason I chose opencv is due to all its capabilities and image processing algorithms. Although I never compared performance intuition is that it would be faster in c++, but again, never measured.
I dont see any significant size increase. It uses opencv as a static library, it takes only the code it needs. For example, using the app from this video, the libnative-lib.so is only 3MB for the arm64-v8a arch
Found package configuration file: C:/tools/OpenCV-android-sdk/sdk/native/jni/OpenCVConfig.cmake but it set OpenCV_FOUND to FALSE so package "OpenCV" is considered to be NOT FOUND. I get this error, how can I fix it? opencv Version 4.5 android Studio Version 4.1
@@adrianm9078 Hi, if you have in CMakeLists.txt file smth like "project("prjName")", try to add all #opencv install stuff {set(OpenCV_STATIC on) set(OpenCV_DIR $ENV{OPENCV_ANDROID}/sdk/native/jni) find_package(OpenCV REQUIRED)} under "project("prjName")"
Sorry. I don't understand why you need to copy the code for converting bitmap to mat (and also from mat to bitmap) and paste it into native-lib.cpp. Because those functions are already provided by jnigraphics, why don't we just invoke them from native-lib.cpp?
Should be possible to combine object detection with tflite and opencv, can check my tutorial: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html As for the counting logic itself you'll have to implement on your own of course. Also worth checking is (the great) MediaPipe box tracking: google.github.io/mediapipe/solutions/box_tracking
Hi, it depends, I did an example of object detection using tensorflow lite, here: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html
Hi, I blogged about it in length here: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html I'll try to make a "code review" video if I'll find the time...
It is fun to watch someone so familiar with the tools they are using :) Thanks for the tutorial
You're very welcome!
Well done! This is very clear and to the point, even with typos that happen to us all. Lol 😂 If I could give you two thumbs up I would. 👍👍 +1 Subscribed!
Thank you!! Happy you liked it.
Please continue this tutorial series, we need it badly
This was not planned to be a series, just how to setup and use opencv on Android... For example, what else would you want to see?
Thank you. This helped a lot! Had to switch to C++ so i can reuse the code on iOS
Great!
Really nice and clean explanation, man. Amazing work. Thank you.
Glad it was helpful!
Excellent tutorial! Thank you so much! I have a question, Would it be possible to detect motion and speed by processing camera input using OpenCv? Does it have the functionality for that?
As an example: your device is capturing the environment while it is inside a moving vehicle. Would it be possible to image process the environtment and determine the vehicle is moving and also gauge its speed? Thank you!
If you can find an algorithm to gauge car speed from a stream of images then doing this with opencv on Android should be possible, I'm not familiar with such an algorithm...
If you want to see how to process camera frames in opencv you can follow my playlist "crossolatform tensorflow lite" I hope to upload soon a video about object detection from the camera stream, which has some part using opencv.
@@TheCodingNotebook That would be great! Yeah any info about how to process camera frames would help. Thank you for the answer. There are a couple of algorithms that can extrapolate speed using parallax. I am hoping OpenCv can provide the pixel tracking that is needed for those algorithms.
Thank you so much for this video, in depth explanation indeed. Just asking would this image processing is better than image processing using java/kotlin in term of performance?
I never tried to do any image processing with java/kotlin, reason I chose opencv is due to all its capabilities and image processing algorithms. Although I never compared performance intuition is that it would be faster in c++, but again, never measured.
wonderful tutorial Huge THANKS
Thx!
it is expecting Sigma to be jobject rather than jfloat. This is throwing error. Can anyone please help...
I really show the entire process on the video, try to clone the code and build it...
using OpenCv makes application size very large . can you compile it for some modules only?
I dont see any significant size increase. It uses opencv as a static library, it takes only the code it needs.
For example, using the app from this video, the libnative-lib.so is only 3MB for the arm64-v8a arch
Found package configuration file:
C:/tools/OpenCV-android-sdk/sdk/native/jni/OpenCVConfig.cmake
but it set OpenCV_FOUND to FALSE so package "OpenCV" is considered to be
NOT FOUND.
I get this error, how can I fix it?
opencv Version 4.5
android Studio Version 4.1
did you fix it? I have the same problem
@vananh199 Check here: (maybe you can contact him thru GH)
github.com/ValYouW/AndroidOpenCVDemo/issues/1
Has anyone fixed it?
@@adrianm9078 Hi, if you have in CMakeLists.txt file smth like "project("prjName")", try to add all #opencv install stuff {set(OpenCV_STATIC on)
set(OpenCV_DIR $ENV{OPENCV_ANDROID}/sdk/native/jni)
find_package(OpenCV REQUIRED)} under "project("prjName")"
After adding the environment variable OPENCV_ANDROID the machine have to be restarted. That might be the cause of the error.
Sorry. I don't understand why you need to copy the code for converting bitmap to mat (and also from mat to bitmap) and paste it into native-lib.cpp.
Because those functions are already provided by jnigraphics, why don't we just invoke them from native-lib.cpp?
Maybe you are right, can you show an example?
Hi buddy good video i wanted to know where you leart all ndk developement as i want to learn but not having much access to study material.
Hi, just thru googling and collecting examples...
Good code from Great man, thank you so much
Nice video, do you know if it's possible to count objects in real time using opencv+object detector model tflite in android?
Should be possible to combine object detection with tflite and opencv, can check my tutorial: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html
As for the counting logic itself you'll have to implement on your own of course.
Also worth checking is (the great) MediaPipe box tracking: google.github.io/mediapipe/solutions/box_tracking
@@TheCodingNotebook Thank you very much!
Nice man, thanks
hello! nice tutorial! do you know how to load dnn models using also c++native code? where to copy the files and how to read them?
Hi, it depends, I did an example of object detection using tensorflow lite, here: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html
Oh, Thank you so much my friend !!!!! very nice tutorial with clear explanation! You saved me a lot of time. Tnx a lot!
Glad to hear that!
היי רציתי לשאול בנוגע לשימוש ב opencv ומצלמה של המכשיר בזמן אמת, נתקלתי בבעיה רצינית אשמח עם אוכל לשאול
Sure, you can also look at this example I did for real-time object detection:
github.com/ValYouW/crossplatform-tflite-object-detecion
Can you make Tensorflow Lite on Android native with C++ video ?
Hi, I blogged about it in length here: www.thecodingnotebook.com/2019/11/cross-platform-object-detection-with.html
I'll try to make a "code review" video if I'll find the time...
@@TheCodingNotebook Thank you very much.
thank you so much
You're welcome!
That's super
Thx!
top
thx
#include not found even after build the project😒
Check the gradle output when doing "gradle sync", sometimes "find_package(OpenCV REQUIRED)" will error and the error message can be easily overlooked