search

stereolabs / zed-yolo

3D Object detection using Yolo and the ZED in Python and C++
https://www.stereolabs.com/
MIT License
159 stars 68 forks source link

Tracking objects detected with YOLO #50

Closed Jordi-Gene-Mola closed 2 years ago

Jordi-Gene-Mola commented 3 years ago

Is it possible to use the object tracking function (integrated in the ZED SDK) to track the objects detected with YOLO?

niemiaszek commented 3 years ago

From ZED SDK 3.6 (released week ago) it is now possible. See release notes and docs page (it's actually pretty well done)

Sujay-Mhaske commented 3 years ago

From ZED SDK 3.6 (released week ago) it is now possible. See release notes and docs page (it's actually pretty well done)

Can you tell me how to do that in python entirely . Please

niemiaszek commented 3 years ago

Sorry, you should refer to this docs page (you can switch between languages, default is c++ but there is also python provided) and API reference on using custom detector. Stereolabs provide this sample on object detection with custom detector, but so far only for c++ as it is a new feature. I'm sorry, but I haven't used this this yet, so I can't share any code from myself.

Sujay-Mhaske commented 3 years ago

Sorry, you should refer to this docs page (you can switch between languages, default is c++ but there is also python provided) and API reference on using custom detector. Stereolabs provide this sample on object detection with custom detector, but so far only for c++ as it is a new feature. I'm sorry, but I haven't used this this yet, so I can't share any code from myself.

Okay I have tried merging the python programs from this https://www.stereolabs.com/docs/object-detection/custom-od/ and zed_tensorflow but i am getting errors , can you tell me how to merge these two programs . I have just 1 month experience on python can you help me??

niemiaszek commented 3 years ago

Im sorry, but I can't provide you working code for this. You should edit capture_thread_func() - follow guide on docs page about the configuration of object detection, then ingest detections from tensorflow into zed, and finally display the 3D content in some viever - this might be the hardest part for you, but you could try to use oglviewer from object tracking example. As you are new in python programming it would be reasonable to wait for python example - looking at how it was done with different new features, it should be available in 3.7 sdk release (2 months). I'm not Stereolabs employee, so don't take my words for granted.

Sujay-Mhaske commented 3 years ago

Okay , Thank you for detailed information. I guess i found to get bounding box on custom data here https://github.com/niemiaszek/zed-yolo/blob/master/zed_python_sample/darknet_zed.py , it is in def get_object_depth(depth, bounds): the thing is i dont know how to compile darknet can you help me do it via anydesk it i am getting dll errors.

niemiaszek commented 3 years ago

I am currently away from home - only have my laptop with me. I have not built darknet with windows for a very long time - last time I did it was 2 years ago. For now I would suggest you to use this installation method from alexeyab darknet repo. It requires you to get some build tools like visual studio compiler (for compiling on windows) and cmake (to set up build to your configuration). Also remember to download opencv binaries, just as specified here. If you get some problems I can try to help you.

Sujay-Mhaske commented 3 years ago

Thank you very much , I will try

github-actions[bot] commented 2 years ago

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment otherwise it will be automatically closed in 5 days