search

david8862 / keras-YOLOv3-model-set

end-to-end YOLOv4/v3/v2 object detection pipeline, implemented on tf.keras with different technologies
MIT License
635 stars 220 forks source link

When I train the model with efficentnet backbone,I met an issue。 #27

Open kz258852 opened 4 years ago

kz258852 commented 4 years ago

When I train the model with efficentnet backbone,The code prompts to download a weight automatically. When I download the weight file into the weight file in advance, the code prompts that the weight downloaded in advance does not conform to a certain format,The code still has to download the weight file itself.

david8862 commented 4 years ago

Could you help to provide the detail error logs?

kz258852 commented 4 years ago

Instructions for updating: Call initializer instance with the dtype argument instead of passing it to the constructor A local file was found, but it seems to be incomplete or outdated because the auto file hash does not match the original value of 345255ed8048c2f22c793070a9c1a130 so we will re-download the data. Downloading data from https://github.com/Callidior/keras-applications/releases/download/efficientnet/efficientnet-b0_weights_tf_dim_ordering_tf_kernels_autoaugment_notop.h5

david8862 commented 4 years ago

from the error info seems the pretrained weights file you provided is not the official one. You can use following official link to get it:

wget https://github.com/Callidior/keras-applications/releases/download/efficientnet/efficientnet-b0_weights_tf_dim_ordering_tf_kernels_autoaugment_notop.h5

And put it under keras model cache path(for ubuntu OS it should be ~/.keras/models/)

kz258852 commented 4 years ago

OK ，Thank you,I will try it 谢谢啊

david8862 commented 4 years ago

不客气 :)