🌈📸 OBF-Design

This framework learns optimal Gaussian Optical Bandpass Filters that transform high-dimensional spectral signatures into compact, information-rich representations 🌟

Deep Gaussian Optical Bandpass Filter Design for Fermentation Index Estimation in Cocoa Beans 🔬💫

A powerful deep learning framework for learnable Optical Bandpass Filter Design! 🎯 Transform spectral signatures into meaningful insights with AI-optimized Gaussian filters 📊✨

🎨 Core Innovation

FilterDesign is the ⭐ core ⭐ of this work!
It learns Gaussian optical filters that reduce spectral dimensionality while preserving the most discriminative features 🌟

🔬 Technical approach:

• 📡 Processes raw spectral data (hundreds to thousands of bands)
• 🧠 Learns optimal Gaussian filter parameters (μ, σ)
• ✂️ Reduces dimensionality to physically feasible spectral bands
• 🎯 Achieves state-of-the-art performance with significantly fewer spectral inputs ✨

🚀 Quick Start

🏃‍♀️ Run FilterDesign with 6 Gaussian filters

python deep_learning.py --mode filter_design --learned-bands 6 --epochs 10

🔄 Compare with baseline (all spectral bands, no filtering)

python deep_learning.py --mode baseline --epochs 10

🎲 Experiment with different filter counts

python run.py  # Tests 3, 6, 11 filters automatically

🛠️ Available Methods

Mode	Description
filter_design	🌟 Primary method - Learnable Gaussian optical filters
band_selection	Learnable binary spectral band selection
binary_band_selection	Hard binary spectral band selection
baseline	Standard full-spectrum approach (no filtering)

🏗️ Technical Architecture

• 🧪 FilterDesign: Gaussian optical filters with learnable μ & σ
• 🤖 Multiple Backbones: SpectraNet, CNN, LSTM, Transformer, SpectralFormer
• 📈 Two-Stage Training: Joint optimization + filter parameter freezing
• 🎛️ FWHM Constraints: Bandwidth limited to physically feasible range (8.25–41.25 nm)

📊 Performance Benefits

• 🎯 Spectral Efficiency: Drastically reduces dimensionality with minimal performance loss
• 💡 Interpretability: Highlights relevant spectral regions linked to the task
• 🔧 Modularity: Compatible with diverse neural architectures
• 📈 Validated: Demonstrated superiority over state-of-the-art band selection methods

⭐ Citation

If you find the OBF-Design useful in your research, please consider citing:

@INPROCEEDINGS{11156301,
  author={Diaz-Delgado, Laura C. and Monroy, Brayan and Bacca, Jorge and Arguello, Henry},
  booktitle={2025 XXV Symposium of Image, Signal Processing, and Artificial Vision (STSIVA)}, 
  title={Deep Gaussian Optical Bandpass Filter Design for Fermentation Index Estimation in Cocoa Beans}, 
  year={2025},
  volume={},
  number={},
  pages={1-5},
  keywords={Optical filters;Filters;Biomedical optical imaging;Optical design;Estimation;Optical computing;Optical imaging;Fermentation;Optical sensors;Optical signal processing;fermentation;cocoa;deep learning;spectral imaging;filter design},
  doi={10.1109/STSIVA66383.2025.11156301}}

Bringing the magic of learnable optics to spectral analysis 💖

🌸 Keep learning, keep filtering! 🌸