Modernize Codebase for Better Compatibility

[Rajdeep-naha]

1. NumPy Compatibility Updates

• Replaced all deprecated NumPy types with modern equivalents:
  • np.bool → bool or np.bool_
  • np.int → int or np.int64
  • np.float → float or np.float64
• Updated type hints and documentation to reflect these changes

2. C++/CUDA KNN Extension Improvements

• Cross-Platform Compatibility:
  • Replaced long with int64_t in all C++/CUDA files
  • Added #include <cstdint> for fixed-width integer types
  • Updated function signatures in:
    • knn.h
    • cpu/vision.h
    • cuda/vision.h
    • cpu/knn_cpu.cpp
    • cuda/knn.cu

3. CLI and Error Handling

• demo.py:
  • Added comprehensive argparse for flexible command-line arguments
  • Improved error messages for missing files/invalid inputs
  • Added progress indicators for long-running operations
• graspnet_dataset.py:
  • Enhanced file path handling with os.path.join()
  • Added validation for input files and data integrity

4. File Path and Cross-Platform Fixes

• Replaced hardcoded path separators with os.path.join()
• Added proper path handling in:
  • data_utils.py
  • graspnet_dataset.py
  • demo.py

5. Code Quality and Maintenance

• New .gitignore:
  • Added comprehensive patterns for Python, build artifacts, IDEs
  • Excluded compiled extensions and temporary files
• Improved Documentation:
  • Added/updated docstrings
  • Clarified function purposes and parameters

6. Testing and Verification

• Successfully built KNN extensions on Windows
• Verified demo script runs without errors
• Confirmed basic functionality with sample data

Files Changed

1. Python Files:

   • demo.py
   • dataset/graspnet_dataset.py
   • utils/data_utils.py
   • utils/loss_utils.py

2. C++/CUDA Files:

   • knn/src/knn.h
   • knn/src/cpu/vision.h
   • knn/src/cpu/knn_cpu.cpp
   • knn/src/cuda/vision.h
   • knn/src/cuda/knn.cu

3. Configuration:

   • .gitignore

Notes for Reviewers

• All changes are backward compatible
• No changes to core functionality