search

AlvaroCavalcante / auto_annotate

Labeling is boring. Use this tool to speed up your next object detection project!
https://medium.com/p/acf410a600b8#9e0e-aaa30a9f4b7a
Apache License 2.0
155 stars 33 forks source link

How to get pretrain model? #13

Closed Laudarisd closed 3 years ago

Laudarisd commented 3 years ago

Hello Alvaro, thanks for the nice tool. I am trying to use your idea.

Would you kindly let me know the way to get pretrained model? Thanks.

One more thing, so tf model folder and auto_annotate stays separately of do i have to put tf model inside of auto_annotate?

Laudarisd commented 3 years ago

I think I solved this problem. Thanks

AlvaroCavalcante commented 3 years ago

Ok, If you need help, just open another issue, I'll be glad to help you! :)

Vibha111094 commented 3 years ago

Hi, I would be really interested to know how to get the pre-trained model. Is it like you label a few images and fine-tune them on an object detection model like YOLO?

Laudarisd commented 3 years ago

@Vibha111094 In my understanding, pre trained models are easily available in open sources. Check here : https://github.com/tensorflow/tfjs-models.

If you are thinking to make your own pre trained model them then its going to be time consuming and challenging.

This might give you simple idea about making resnet model in keras: https://towardsdatascience.com/hitchhikers-guide-to-residual-networks-resnet-in-keras-385ec01ec8ff

Even training YOLO usages pre trained mode. I hope I answered your question.

Vibha111094 commented 3 years ago

@Laudarisd thankyou for your reply. However, I am not very clear what is the data that the pre-trained model is trained on (for this particular use case)