yanglf1121/nuset

NuSeT: A Deep Learning Tool for Reliably Separating and Analyzing Crowded Cells

nuset's Issues

NotImplementedError: Cannot convert a symbolic Tensor (meshgrid/Size_1:0) to a numpy array

I installed recommended dependencies, but got NotImplementedError: Cannot convert a symbolic Tensor (meshgrid/Size_1:0) to a numpy array
here is my conda environment, OS is Linux Mint 20.2

name: nuset
channels:
  - defaults
  - conda-forge
  - bioconda
dependencies:
  - python=3.7
  - pillow
  - scikit-image
  - tensorflow-gpu==1.15.0
  - numpy
  - scikit-image
  - tqdm

Training label generation

Can you tell us about the exact nature of the training labels used? Do the masks for different cells need to be of a different color or have to be of the same color? Should the background be black or white? Thank you!

Window wont show up

Hello,
I install the requirements by following the instructions in your repo, and when I try to run the program by python NuSet.py in terminal, I get the following message and that is the only message I get, the window wont show up .
I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_100.dll
Could you please help me in this issue?

Update TF 1.15 to TF 2.3?

I am looking to integrate Nuset into some of my projects. My models are trained on tf 2.3 and I'll be using Nuset as sort of a pre-step to my models. It's a bit cumbersome to build a pipeline that switches between tf1.15 and tf 2.3.

Hence, is it possible to bump nuset to use tf 2.3 only?