predict 10~100 bbxes in feature extraction

airsplay / py-bottom-up-attention

PyTorch bottom-up attention with Detectron2

Apache License 2.0

229 stars 56 forks source link

predict 10~100 bbxes in feature extraction #23

Open RubickH opened 3 years ago

RubickH commented 3 years ago

Thanks for your codes and models! I notice that you have provided the model trained with 10-100 bbxes. I wonder how I can use this model to predict 10-100 bbxes in feature extraction. It seems that the number of features is closely related to the hyperparameter 'NUM_OBJECTS'. I don't know how to set it in the adaptive feature extraction mode. Thanks!