Add Shape inference

[alifahrri]

Type of changes

Please check the type of change your PR introduces: - [x] Bugfix - [ ] Feature - [ ] Code style update (formatting, renaming) - [ ] Refactoring (no functional changes, no api changes) - [ ] Build related changes - [ ] Documentation content changes - [ ] Other (please describe): ## What is the current behavior? no shape inference - close #39 Related issues : ## What is the new behavior?

• add shape inference
• add tensorrt docker
• add model explorer

Checklist

• [ ] Unit tests pass locally with my changes
• [ ] I have added tests that prove my fix is effective or that my feature works
• [x] I have added necessary documentation (if appropriate)
• [x] Any dependent changes have been merged and published in downstream modules
• [x] I have filled the [Unreleased] Changelog

[alifahrri]

Note:

• the tensorrt docker is adapted from github.com/microsoft/onnxruntime/, and use python3.7
• for now only vortex runtime available from tensorrt docker
• known dependency issues: albumentations, enforce
• model explorer provided also measure inference time, provided as alternative to model reporting
• several helper class in the examples are added since vortex development cant be installed yet, suggestion on where to put those classes are very much appreciated
• there are some errors when testing, but i think it is unrelated to this pr

  
  def _conv_forward(self, input, weight):
      if self.padding_mode != 'zeros':
          return F.conv2d(F.pad(input, self._reversed_padding_repeated_twice, mode=self.padding_mode),
                          weight, self.bias, self.stride,
                          _pair(0), self.dilation, self.groups)
      return F.conv2d(input, weight, self.bias, self.stride,
  >                       self.padding, self.dilation, self.groups)
  E       RuntimeError: Given groups=1, weight of size [256, 144, 1, 1], expected input[1, 352, 26, 26] to have 144 channels, but got 352 channels instead

========================================================================================= short test summary info ========================================================================================== FAILED tests/test_model.py::test_model[detection-darknet7] - RuntimeError: Sizes of tensors must match except in dimension 1. Got 40 and 20 in dimension 2 (The offending index is 1) FAILED tests/test_model.py::test_model[detection-efficientnet_b0] - RuntimeError: Given groups=1, weight of size [256, 192, 1, 1], expected input[1, 368, 40, 40] to have 192 channels, but got 368 chann... FAILED tests/test_model.py::test_model[detection-mobilenet_v2] - RuntimeError: Given groups=1, weight of size [256, 176, 1, 1], expected input[1, 352, 40, 40] to have 176 channels, but got 352 channels...

[triwahyuu]

I think you forgot to add sympy in the requirements

[triwahyuu]

I have fixed the failed tests in #85

[alifahrri]

I think you forgot to add sympy in the requirements

I think you forgot to add sympy in the requirements

fixed

[alifahrri]

can you elaborate what the scripts/model_explorer.py does? I tried to use it but I don't understand how to do anything with it.

model_explorer is similar to dataset visualizer but intended for trying model inference interactively

[triwahyuu]

how can I use it? after I load the model, I dont know what to do

[triwahyuu]

ahh, I got it now, all good. LGTM.