Notifications
Clear all

Question Custom Object Detection

5 Posts
2 Users
0 Likes
399 Views
(@longquan)
Active Member Member
Joined: 8 months ago
Posts: 3
Topic starter  

I have gone through the tutorial for training object classification model and I want to create a custom object detection model using peeking duck. Any tutorial recommendation please? 

I tried to follow this tutorial ( removed link ) and I get a tensorflow lite model. Does this compatible with peeking duck? If yes, how?

Appreciate your help, Thanks!


   
Quote
Gao Hongnan
(@gao-hongnan)
Eminent Member Moderator
Joined: 1 year ago
Posts: 28
 

Hi @longquan, I am unable to view the link, could you post it again? It also depends on which object detection model you are using.


   
ReplyQuote
(@longquan)
Active Member Member
Joined: 8 months ago
Posts: 3
Topic starter  

Thanks for your reply. I am not sure why the link got deleted automatically. Google search 'Model Maker Object Detection for Android Figurine' , it should be the first result, colab notebook. Sorry for the inconvenience.


   
ReplyQuote
Gao Hongnan
(@gao-hongnan)
Eminent Member Moderator
Joined: 1 year ago
Posts: 28
 

Hi thanks for the reply, are you looking for a specific model for your use case? I'm afraid the tutorial you linked me uses a lite version of EfficientDet, whereas we carry the non-lite versions. You may need to create your own inference node to interface with PeekingDuck. In other words, after you trained the model, you obtain a set of weights, after which you will load the weights. Once this step is done, you can write a script (a custom node) to then interface with your outputs.


   
ReplyQuote
(@longquan)
Active Member Member
Joined: 8 months ago
Posts: 3
Topic starter  

Alright Thank you! Let me get back you if I face any problem.


   
ReplyQuote
Share: