search

anilsathyan7 / Portrait-Segmentation

Real-time portrait segmentation for mobile devices
MIT License
638 stars 133 forks source link

Need help to change input shape. #48

Closed ELTON-LGB closed 3 years ago

ELTON-LGB commented 3 years ago

Thanks for this awesome repository! Thanks for your great researching!

I passed training for slim512 net, and got what I want. My question is:

  1. How can I change the input tensor shape to [256, 256, 3]? I changed the Input to [256, 256,3], but failed, I got error: ValueError: Operands could not be broadcast together with shapes (32, 32, 128) (16, 16, 128), so it is not easy only to change Input shapes and image_size.

  2. "The inputs are initially downsampled from a size of 512 to 128 (i,e 1/4'th)", If changed input tensor shape to [256,256,3], can save reference time with same accuracy? or Not?

  3. I want to keep the current accuracy, and save more reference time, how can I improve the slim-net? change input shape can do it?

Thanks again for your great works!

anilsathyan7 commented 3 years ago

You may lose accuracy if you reduce input size; even though it may give you a slight speed-up. The error could be due to the resize values hardcoded into the decoder layers in the model definition.

ELTON-LGB commented 3 years ago

Thanks so much!