Open3D-ML Jupyter Tutorials #512
Conversation
* Fixed code errors in train_ss_model_using_pytorch.ipynb * Fixed code errors in train_ss_model_using_tensorflow.ipynb * Updated content in all tutorials
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
- This needs an introduction at the top describing what we will do in this tutorial.
- The language describing the steps is not quite right. @sanskar, you may need to directly edit Alex's PR. e.g: (-in our data model, we define a dataset....)
- The note at the top about downloading weights is specific to PyTorch. Add TF alternative.
Reply via ReviewNB
There was a problem hiding this comment.
To Do (Sanskar): editing pass
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
Add (commented) command to download the weights directly in the notebook:
# from urllib.request import urlretrieve
# urlretrieve(weights_url, filename=weights_file)
Reply via ReviewNB
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,608 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,608 @@ | |||
| { | |||
There was a problem hiding this comment.
Checking the type is not useful. Instead use
vars(cfg)
This will list the 3 dictionaries in cfg
Reply via ReviewNB
| @@ -0,0 +1,608 @@ | |||
| { | |||
There was a problem hiding this comment.
Mention that items can be viewed as well as updated like in a standard Python dict.
Reply via ReviewNB
| @@ -0,0 +1,373 @@ | |||
| { | |||
There was a problem hiding this comment.
Only keep the part of the image inside the red box - that's clearer than the entire image + arrow.
Reply via ReviewNB
| @@ -0,0 +1,373 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
This file needs an editing pass for language and organization. Also, rephrase from a user's perspective who is trying to learn what functionality is available and exactly what it will do. As an example, mention the inputs needed to make the main steps work (e.g. how do you get the dataset?) and what is the result. e.g. after running training, you should have a checkpoint (where?) that you can use for testing or inference.
Reply via ReviewNB
There was a problem hiding this comment.
To Do (Sanskar): editing pass for language and organization
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
In this tutorial, we will learn how to train a semantic segmentation model using PyTorch in a Jupyter Notebook. We assume that you are familiar with Jupyter Notebook and have created a folder notebooks in a folder that is relative to ml3d.
Reply via ReviewNB
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
Needs editing to remove repeating language:
Run inference with a pre-trained model. For this example, we will use the RandLANet model.Rephrase:
Train a model. We will train a model using theSemanticKITTIdataset andRandLANetmodel.
Rephrase for clarity:
Run an inference and run a test. We will run an inference using the 'training' split that use a pointcloud and display a result. However, a test is run on a pre-defined test set rather than a pass pointcloud.
Reply via ReviewNB
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
For this example, we will use the RandLANet model.This is already covered earlier. Avoid repetition.
Reply via ReviewNB
| @@ -0,0 +1,333 @@ | |||
| { | |||
There was a problem hiding this comment.
Running a test is very similar to training the model.
Instead, say what this will do:
Next we will evaluate the trained model on the test split.
Reply via ReviewNB
for * Reading a Config File * Training Semantic Segmentation Model Using PyTorch
Sanskar, Sameer, and Benjamin,
This PR is for newly created Open3D-ML Jupyter Tutorials. They were run from the
Open3D-ML/docs/tutorial/notebook/path.Please feel free to critique or suggest changes or improvements.
Thank you all very much!
This change is