PRBonn / make_it_dense

Make it Dense: Self-Supervised Geometric Scan Completion of Sparse 3D LiDAR Scans in Large Outdoor Environments
https://www.ipb.uni-bonn.de/wp-content/papercite-data/pdf/vizzo2022ral-iros.pdf
133 stars 12 forks source link

Pull Request: Updates to `train.py` and `data.py` #10

Closed minwoo0611 closed 11 months ago

minwoo0611 commented 11 months ago

This pull request introduces two significant changes:

  1. Modification in apps/train.py:

    • Context: Compatibility issues with the latest version of Torch Lightning.
    • Changes Made:
      • Updated the Train class to accommodate changes in variable handling due to the updated torch lightning version.
      • Adjusted parameters including gpus and weights_summary.
      • Note: The parameter auto_scale_batch_size remains unchanged.
  2. Update in src/make_it_dense/utils/data.py:

    • Issue: Encountered a ValueError related to the random.sample() method, indicating a sample size larger than the population or negative.
    • Resolution: Replaced random.sample() with random.choices() to prevent this error and ensure more robust sampling, especially in cases with smaller datasets.

These changes aim to enhance compatibility with newer library versions and improve the robustness of the data handling process.

nachovizzo commented 11 months ago

Thanks!