Convert.py does not produce yolov3.tf

Question

Convert.py does not produce yolov3.tf

kalikhademi opened this issue 3 years ago · comments

Kiana Alikhademi, Ph.D. commented 3 years ago

Hi

I am running convert.py on darknet original weights. It does not produce any error but it does not produce yolov3.tf although I put it as the output file. It produces the following files.

I appreciate if anyone can help me about this.

Taru Rustagi · Answer 1 · Thu Jun 24 2021 11:54:25 GMT+0800 (China Standard Time)

It doesn't matter if the extensions are not the same. Run the inference still with 'yolov3.tf', and it correctly reads the data automatically.

Kiana Alikhademi, Ph.D. · Answer 2 · Fri Jun 25 2021 00:04:20 GMT+0800 (China Standard Time)

Thanks! I used the .tf in the detect.py and it worked but in the training .tf was not working.

rajan780 · Answer 3 · Sun Jul 18 2021 02:01:02 GMT+0800 (China Standard Time)

Hi @kalikhademi
In train.py if you are loading weights from .tf file please replace the model.load_weights(FLAGS.weights) with model.load_weights(FLAGS.weights).expect_partial()