Plastic and Filler Classification and Regression Tool
AI-assisted FTIR spectra classification for plastic composition analysis and regression for filler estimation in PP
From FTIR Spectra to Plastic and Filler Information
This tool supports the characterisation of mixed plastic streams using Fourier-transform infrared (FTIR) spectroscopy and deep learning. Users can upload FTIR spectra from plastic samples, which are then automatically checked, pre-processed, and analysed by pre-trained AI models.
The first model defines the plastic type and provides a confidence score for each prediction. The classification model is trained on over 2,800 waste samples manually cross-checked for PP, PE, ABS, HIPS, PC/ABS, PA, or PVC. For samples classified as PP, a second pre-trained regression model estimates talc, calcium carbonate, and total filler contents with an error below ±3 wt%. When matching XRF results are uploaded, bromine (Br), antimony (Sb), and chlorine (Cl) values can be merged into the final output table using filename-based matching.
For questions about the FTIR Plastic and Filler Analysis Tool, contact ftir@sude.be.
Step 1: Choose how to upload FTIR spectra
Step 2: Confirm XRF option and download the matching example
Step 3: Upload your files and run analysis
No FTIR workbook selected | XRF: none
Results will appear here after uploading the FTIR spectra (optionally together with XRF results).
This tool was further developed within the SERENADE and INCREACE projects, both funded by the European Union. SERENADE is funded under the Horizon Europe MSCA Doctoral Networks programme, Grant Agreement No. 101072846. INCREACE is funded under the Horizon Europe research and innovation programme, Grant Agreement No. 101058487. serenade-project.eu