feature-extraction-and-classification

This project enables image classification based on feature extraction.

Project Structure

manufacturing-part-classification/
│
│
├── .github/                     # GitHub specific files like workflows for CI/CD
│   └── workflows/
│
├── .vscode/                     # VSCode specific settings, including debug configurations
│   └── launch.json
│
├── docs/                        # Sphinx documentation source files
│
├── src/                         # Source code for the project
│   ├── config/                  # Configuration files for all type of models
│   ├── dataloader/              # DataLoader implementation for the project
│   ├── model/                   # Model definitions and utilities
│   ├── scripts/                 # Scripts for training, evaluation, etc.
│   ├── utils/                   # Utility scripts for logging, data manipulation, etc.
│   └── launch.py                # Main script to launch training with configuration
│
├── .gitignore                   # Specifies intentionally untracked files to ignore
├── LICENSE                      # License file
└── README.md                    # Project overview and setup instructions

Pipeline Overview

The pipeline is structured as follows:

• Preprocess: This step includes all necessary data transformation activities to ready the data for feature extraction.
• Feature Extraction: This involves the extraction of pertinent features from the preprocessed images, utilizing various methods like SIFT or VGG16.
• Feature Prediction: At this phase, the features that have been extracted are utilized to predict the categories of the manufacturing parts. The models used here may include K-NN, SVC, or RandomForest.

Installation

# Create Environment
conda create -n <env_name> python==3.9

#Activate environment
conda activate <env_name>

#Install dependencies
pip install numpy PyYAML opencv-python scikit-learn tensorflow==2.9.0

#Optional for generating documentation
pip install sphinx sphinx_rtd_theme

Quick Start

Training and Validation

cd ./src

python ./launch.py --config <config_file>

#Example
python ./launch.py --config ./config/vgg16.yaml

Testing

python ./launch.py --config ./config/vgg16.yaml --test-only --weights <weights_file>

Feature Visualization

python ./launch.py --config <config_file> --visualization-only --weights <weights_file>

Model Zoo

Feature Extractor	Link
SIFTFeatureExtractor	class
VGG16FeatureExtractor	class

Feature Predictor	Link
RandomForestPredictor	class
SVMPredictor	class
KNNPredictor	class

Citation

If you find Manufacturing-Part-CLassification useful, please cite my work as an acknowledgment.

@misc{feature-extraction-classification2024,
    title={Feature Extraction and Classification},
    author={Aleman Mihnea},
    howpublished={\url{https://github.com/SpeedyGonzales949/feature-extraction-classification}},
    year={2024}
}