README.md

metadata

title: AI Polymer Classification (Raman & FTIR)
emoji: 🔬
colorFrom: indigo
colorTo: yellow
sdk: streamlit
app_file: Upload_and_Run.py
pinned: false
license: apache-2.0

AI-Driven Polymer Aging Prediction and Classification (v0.1)

This web application classifies the degradation state of polymers using Raman and FTIR spectroscopy and deep learning. It is a prototype pipeline for evaluating multiple convolutional neural networks (CNNs) on spectral data.

---

🧪 Current Scope

• 🔬 Modalities: Raman & FTIR spectroscopy
• 💾 Input Formats: .txt, .csv, .json (with auto-detection)
• 🧠 Models: Figure2CNN (baseline), ResNet1D, ResNet18Vision, Custom CNNs (Enhanced, Efficient, Hybrid)
• 📊 Task: Binary classification — Stable vs Weathered polymers
• 🚀 Features:
  • Single-spectrum + Batch Spectrum Analysis
  • Multi-model comparison
  • Performance tracking dashboard
• 🛠️ Architecture: PyTorch + Streamlit

---

🚧 Roadmap

• Inference from Raman .txt files
• Model selection (Figure2CNN, ResNet1D)
• FTIR support (modular integration complete)
• Multi-model comparison dashboard
• Performance tracking dashboard
• Add more trained CNNs for comparison
• Image-based inference (future modality)
• RESTful API for programmatic access

---

🧭 How to Use

The application provides three main analysis modes in a tabbed interface:

1. Standard Analysis:

   • Upload a single spectrum file (.txt, .csv, .json) or a batch of files.
   • Choose a model from the sidebar.
   • Run analysis and view the prediction, confidence, and technical details.

2. Model Comparison:

   • Upload a single spectrum file.
   • The app runs inference with all available models.
   • View a side-by-side comparison of the models' predictions and performance.

3. Performance Tracking:

   • Explore a dashboard with visualizations of historical performance data.
   • Compare model performance across different metrics.
   • Export performance data in CSV or JSON format.

Supported Input

• Plaintext .txt, .csv, or .json files.
• Data can be space-, comma-, or tab-separated.
• Comment lines (#, %) are ignored.
• The app automatically detects the file format and resamples the data to a standard length.

---

Contributors

• Dr. Sanmukh Kuppannagari (Mentor)
• Dr. Metin Karailyan (Mentor)
• Jaser Hasan (Author/Developer)

Model Credit

Baseline model inspired by:

Neo, E.R.K., Low, J.S.C., Goodship, V., Debattista, K. (2023). Deep learning for chemometric analysis of plastic spectral data from infrared and Raman databases. Resources, Conservation & Recycling, 188, 106718. https://doi.org/10.1016/j.resconrec.2022.106718

---

🔗 Links

• Live App: Hugging Face Space
• GitHub Repo: ml-polymer-recycling

---

🚀 Technical Architecture

The system is built on a modular, production-ready architecture designed for scalability and maintainability.

• Frontend: Streamlit-based web application (Upload_and_Run.py) with interactive, multi-tab UI.
• Backend: PyTorch for deep learning operations including model loading and inference.
• Model Management: Registry pattern (models/registry.py) for dynamic model loading and easy integration of new architectures.
• Data Processing: Modality-aware preprocessing pipeline (utils/preprocessing.py) for data integrity and standardization (Raman & FTIR).
• Multi-Format Parsing: utils/multifile.py for parsing .txt, .csv, and .json files.
• Results Management: utils/results_manager.py for managing session and persistent results, multi-model comparison, and data export.
• Performance Tracking: utils/performance_tracker.py logs metrics to SQLite and powers the dashboard.
• Deployment: Containerized via Docker (Dockerfile) for reproducible, cross-platform execution.

---

Notable Additions in space-deploy Branch

• Enhanced FTIR support, with modular integration for spectral data.
• Multi-model comparison dashboard for evaluating multiple CNNs in parallel.
• Performance tracking dashboard with export options (CSV, JSON).
• Batch spectrum analysis for high-throughput evaluation.
• Updated color scheme (teal → yellow) for improved UI clarity.
• Improved README with Hugging Face Space config and clearer usage instructions.
• Bug fixes for filename casing and SDK deployment settings.
• Forward-looking plans for RESTful API and image-based inference.

---