HMB Helpers Package

A comprehensive collection of helper modules for image processing, segmentation, deep learning workflows, text/PDF utilities, and scientific computing in PyTorch, TensorFlow, and beyond.

Documentation: https://hmb-helpers-package.readthedocs.io/en/latest/

---

Table of Contents

• Motivation
• Installation
• Dependencies
• Features & Modules
• Documentation
• Testing
• Contributing
• Citation & License
• Support & Contact

---

Motivation

HMB Helpers Package aims to accelerate research and development in computer vision, deep learning, and text analytics by providing ready-to-use, well-tested utility modules that simplify common tasks, reduce boilerplate code, and promote reproducibility in scientific projects.

Installation

Minimal Install (Recommended)

Install only core dependencies (numpy, pillow):

pip install hmb-helpers

Package on PyPI: https://pypi.org/project/hmb-helpers/

Install with Optional Features

Add only the features you need:

# Computer vision & PyTorch.
pip install "hmb-helpers[cv,pytorch]"

# NLP & text processing.
pip install "hmb-helpers[nlp]"

# PDF handling.
pip install "hmb-helpers[pdf]"

# Full installation (most optional dependencies)
# Note: The `all` extra installs most optional dependencies but intentionally
# excludes large, platform- and device-specific frameworks (PyTorch, TensorFlow,
# Keras and related runtime packages). Install those frameworks separately
# using the dedicated extras below or via the framework's official installer.
pip install "hmb-helpers[all]"

Development Install

For modifying the package source:

git clone https://github.com/HossamBalaha/HMB-Helpers-Package.git
cd HMB-Helpers-Package
pip install -e ".[dev]"

GPU Support (PyTorch CUDA)

The pytorch extra installs CPU wheels by default. For CUDA:

pip install "hmb-helpers[pytorch]"
pip uninstall torch torchvision torchaudio -y
pip install torch==2.7.1+cu128 torchvision==0.22.1+cu128 torchaudio==2.7.1+cu128 \
  --extra-index-url https://download.pytorch.org/whl/cu128

Or use the official PyTorch installer for your platform.

Dependencies

Core Dependencies (Always Installed)

• numpy>=1.26.4,<2: Numerical computing
• pillow>=12.2.0: Image I/O and basic processing
• pyyaml>=6.0.3: YAML parsing for configuration and dataset helper utilities
• pandas>=3.0.2: Tabular data handling used across multiple helpers
• matplotlib>=3.9: Basic plotting utilities used by helpers and examples
• tqdm>=4.67.3: Progress bars used in many processing functions
• scikit-learn>=1.8.0: Common ML utilities (encoders, imputers, scalers) used by preprocessors

Optional Feature Groups

Install only what you need via extras:

Feature	Command	Key Packages
Scientific Stack	[scientific]	scipy, pandas, scikit-learn, scikit-image
Computer Vision	[cv]	opencv-python, imagehash, pyvips
PyTorch	[pytorch]	torch, torchvision, torchaudio
TensorFlow	[tensorflow]	tensorflow, keras, tf-keras
NLP	[nlp]	nltk, spacy, transformers, gensim
PDF	[pdf]	PyMuPDF, PyPDF2, tabula-py
Audio	[audio]	librosa, spafe, praat-parselmouth
Medical Imaging	[medical]	pydicom, nibabel, openslide-python
Classical ML	[ml]	xgboost, catboost, lightgbm, optuna
Visualization	[plotting]	matplotlib, seaborn, plotly
Utilities	[utils]	tqdm, albumentations, shap, trimesh

See requirements.txt for exact version pins used in development. Note that pip install "hmb-helpers[all]" purposefully omits PyTorch/TensorFlow/Keras so that users can install the appropriate platform-specific (CPU/CUDA) wheels via the framework vendor instructions or by selecting the framework extras explicitly (for example pip install "hmb-helpers[pytorch]").

Features & Modules

Core Modules

1. AgentsHelper: AI agent orchestration and interaction utilities.
2. ArabicTextHelper: Specialized tools for Arabic text processing and analysis.
3. AttentionMapsHelper: Tools for generating and visualizing attention maps in deep learning models.
4. AudioHelper: Audio processing, feature extraction, and manipulation utilities.
5. CompressionsHelper: Data compression and decompression utilities.
6. DataAugmentationHelper: Image and data augmentation pipelines.
7. DatasetsHelper: Utilities to detect, prepare, and validate image classification datasets (train/val/test layouts).
8. EmbeddingsToTextHelper: Convert between embeddings and text representations for NLP tasks.
9. ExplainabilityHelper: Model explainability and interpretability (e.g., SHAP analysis).
10. HandCraftedFeatures: Feature extraction utilities for images and tabular data.
11. ImagesComparisonMetrics: Image comparison metrics (SSIM, PSNR, MSE, etc.).
12. ImageSegmentationMetrics: Segmentation evaluation metrics (IoU, Dice, pixel accuracy).
13. ImagesHelper: Comprehensive image loading, saving, resizing, cropping, and manipulation.
14. ImagesNormalization: Image normalization, standardization, and color space conversion.
15. ImagesToEmbeddings: Extract embeddings from images using timm and transformers models.
16. Initializations: Model and layer initialization helpers for deep learning frameworks.
17. MachineLearningHelper: ML workflow helpers (data splitting, cross-validation, model selection).
18. MetaheuristicsHelper: Metaheuristic optimization algorithms (e.g., MRFO).
19. PDFHelper: PDF reading, extraction, manipulation, and annotation.
20. PerformanceMetrics: Comprehensive performance metrics for classification and regression.
21. PlotsHelper: Plotting and visualization helpers (wrappers around matplotlib/seaborn utilities).
22. PyTorchClassificationLosses: Custom classification loss functions for PyTorch.
23. PyTorchHelper: PyTorch utilities for models, tensors, device management, and checkpointing.
24. PyTorchModelMemoryProfiler: Utilities to profile PyTorch model memory usage.
25. PyTorchSegmentationLosses: Custom segmentation losses (Dice, BCE, DiceBCE, Focal, Tversky, IoU).
26. PyTorchTabularModelsZoo: Collection of tabular model utilities and reference architectures.
27. PyTorchTrainingPipeline: Training pipeline helpers for PyTorch experiments (data loaders, trainers, schedulers).
28. PyTorchUNetModelsZoo: UNet architecture implementations and utilities for PyTorch.
29. StatisticalAnalysisHelper: Statistical analysis and data exploration tools.
30. StringsHelper: String manipulation and text processing utilities.
31. TextGenerationMetrics: Metrics for text generation models (ROUGE, BLEU, METEOR).
32. TextHelper: Text normalization, cleaning, tokenization, and NLP utilities.
33. TFAttentionBlocks: TensorFlow/Keras attention mechanism implementations.
34. TFHelper: TensorFlow/Keras utilities and helpers (Grad-CAM, etc.).
35. TFSegmentationLosses: TensorFlow/Keras segmentation loss implementations.
36. TFUNetHelper: TensorFlow/Keras UNet-related helpers and utilities.
37. Utils: Miscellaneous utilities for file I/O, configuration, and data handling.
38. VectorsHelper: Vector operations and geometric computations.
39. VideosHelper: Video processing and frame extraction utilities.
40. VotingHelper: Ensemble voting methods for machine learning.
41. WSIHelper: Whole Slide Image (WSI) processing for digital pathology.
42. YOLOHelper: YOLO model training and inference utilities.

Documentation

Full documentation is available in the build/html/ directory after building with Sphinx.

On POSIX systems with make available (Linux, macOS):

cd source
make html

Examples

All example scripts, detailed usage, CLI tables and troubleshooting tips live in the examples README:

• HMB/Examples/README.md — beginner-friendly guide with per-script descriptions, example run commands, and compact CLI tables (recommended starting point).

Quick pointers:

• Examples and helper scripts are located under HMB/Examples.
• Platform wrappers (Windows): HMB/Examples/BAT Files/ — run with call from the repository root.
• Platform wrappers (POSIX): HMB/Examples/SH Files/ — run with bash.

If you want a short list of commonly used examples, open HMB/Examples/README.md — it contains a curated list and copyable run commands for each script.

Or run the equivalent POSIX wrapper (bash):

bash "HMB/Examples/SH Files/Timm_Statistics_Analysis_Ablations.sh"

To run an example Python file directly, quote the path if it contains spaces. Examples:

python "HMB\Examples\Timm_Statistics_Analysis_Ablations.py"

python3 "HMB/Examples/Timm_Statistics_Analysis_Ablations.py"

Testing

The package includes comprehensive unit tests for all modules. Run tests using the provided test runner or with pytest directly.

Run the bundled test runner:

# Run all tests.
python tests/run_tests.py

# Run a specific test file.
python tests/run_tests.py Test_ImagesHelper.py

Run tests with pytest (recommended if you have pytest installed):

pytest -q
# or run a specific test file
pytest -q tests/Test_ImagesHelper.py

Test coverage includes:

• Unit tests for all core modules
• Edge case validation
• Integration tests for complex workflows
• Performance and regression tests

Current test status: 530+ tests passing

Contributing

Contributions are welcome! See CONTRIBUTING.md in the repository root for full contributor guidelines (development setup, testing, linting, PR checklist):

• https://github.com/HossamBalaha/HMB-Helpers-Package/blob/main/CONTRIBUTING.md

Quick summary:

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -am 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Development Guidelines (short)

• Write comprehensive docstrings for public functions and classes
• Add unit tests for new functionality
• Follow PEP 8 style guidelines and run black/flake8/mypy
• Update documentation when adding or changing features
• Ensure all tests pass before submitting a PR

Changelog

All notable changes are recorded in CHANGELOG.md in the repository root. See:

• https://github.com/HossamBalaha/HMB-Helpers-Package/blob/main/CHANGELOG.md

Citation & License

This project is licensed under the MIT License. See the LICENSE file for details.

Full license text is available in the repository: https://github.com/HossamBalaha/HMB-Helpers-Package/blob/main/LICENSE

If you use this package in your research, please cite the relevant modules as described in their headers.

Attribution Requirement

If you use this package in your work, please:

1. Include a copy of the LICENSE file with any distribution
2. Credit the author in documentation or publications:

   @software{balaha_hmb_helpers_2026_030,
     author    = {Balaha, Hossam Magdy},
     title     = {HMB-Helpers-Package: HMB-Helpers-Package v0.3.0},
     year      = {2026},
     publisher = {GitHub},
     month     = jul,
     version   = {v0.3.0},
     url       = {https://github.com/HossamBalaha/HMB-Helpers-Package}
   }
   
   @software{hossam_magdy_balaha_2026_21378674,
      author       = {Hossam Magdy Balaha},
      title        = {HossamBalaha/HMB-Helpers-Package: HMB-Helpers-Package v0.3.0},
      month        = jul,
      year         = 2026,
      publisher    = {Zenodo},
      version      = {v.0.3.0},
      doi          = {10.5281/zenodo.21378674},
      url          = {https://doi.org/10.5281/zenodo.21378674},
    }
   

Support & Contact

For questions, bug reports, or contributions, please contact the author:

• Hossam Magdy Balaha
• Email: h3ossam@gmail.com
• Or open an issue on GitHub

---

Happy coding!