search

AlbertoSabater / Robust-and-efficient-post-processing-for-video-object-detection

GNU General Public License v3.0
146 stars 20 forks source link

What is your Keras and Tensorflow version? #22

Closed wjtan99 closed 2 years ago

wjtan99 commented 2 years ago

Thanks for sharing your great work. I try to reproduce your embedding part. What is your Keras and TF version?

AlbertoSabater commented 2 years ago

Hi! REPP does not need any TF or Pytorch library, since it just uses scikit-learn for the matching algorithms. The ML library is dependent on the model you use to extract the embeddings. In my case of YOLOv3 TF 1.13.1 and Keras 2.2.4.

wjtan99 commented 2 years ago

Yes, I was asking about the YOLOv3 Keras and TF versions. Thanks a lot for your quick reply. I still could not get it run. What Python version do you have?

wjtan99 commented 2 years ago

Never mind. I read in other discussions that the no-embedding (appearance) works as well as with appearance. That is good for me.