search

tensorflow / models

Models and examples built with TensorFlow
Other
77k stars 45.78k forks source link

SSD mobiledet spend more time on testing than mobilenetV2 and V3 #8657

Open gmt710 opened 4 years ago

gmt710 commented 4 years ago

Thank you very much for your contribution to lightweight model research. I've trained the mobilenetV2/V3 and Mobiledet by .config file, and tested them.MobilenetV3 is the fastest. Is Mobiledet slower than mobilenetV2? Hope to receive your reply. image

zhaoxin111 commented 4 years ago

Hi, @gmt710 . Can you run the pretrained mobiledet cpu model provided by TF1 model zoo. I encountered the following error:

tensorflow.python.framework.errors_impl.NotFoundError: Op type not registered 'TFLite_Detection_PostProcess' in binary running on localhost.localdomain. Make sure the Op and Kernel are registered in the binary running in this process. Note that if you are loading a saved graph which used ops from tf.contrib, accessing (e.g.) tf.contrib.resampler should be done before importing the graph, as contrib ops are lazily registered when the module is first accessed.

os: linux python:3.6.11 tensorflow:tf 1.15.3

I think it may be caused by the inconsistency between the version I used to run the model and the version used for model training. Can you give me some suggestions for running the mobiledet cpu model.

Best

gmt710 commented 4 years ago

The environment：

linux
python3.6
tensorflow==v1.12 or 1.13

I've tried gpu version, It's ok. Maybe you need to spend some time modifying the corresponding version of tensorflow. especially mobilenetV2.