search

abbypa / NNProject_DeepMask

Deep Neural Network for object segmentation.
187 stars 60 forks source link

Is EndToEnd.py the entrance for the whole project? #7

Closed JieChen91 closed 8 years ago

JieChen91 commented 8 years ago

Hi, I try to first run EndToEnd.py to see how the project works, but there are many pre-set paths, like Resources and Predictions containing other files that are not originally in the repository. Does that mean I have to generate those paths and files in advance by myself?

For example, to obtain 'Resources/vgg16_graph_weights.h5', I need to run HelperScripts/CreateVggGraphWeights.py. However, I get a bug there saying "'Graph' object has no attribute 'layers'", occurred at line 150 (graph.set_weights(model.get_weights())).

Can anyone help? Thanks~

zhengxuyu commented 8 years ago

Yes, indeed, you need to generate those files before you run EndToEnd.py. And download the Coco dataset as well. For the example, you need to install Keras 0.3.1 to solve this problem You can read the closed issue to get some idea

abbypa commented 8 years ago

That's correct.

In general, you need to-

  1. Install all requirements, as listed in the readme
  2. Download mscoco annotations (see readme link)
  3. Download and convert graph weights with HeplerScripts/CreateVggGraphWeights.py (see readme link)
  4. Create the learning dataset using ExamplesGenerator.py
  5. Run EndToEnd.py :)

And the other issue you described is indeed a keras version issue.