Savvas Sifnaios
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Savvas Sifnaios has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
Angelos Amditis
4 shared publicationsGeorgios Tsimiklis
4 shared publicationsFotios K. Konstantinidis
3 shared publications- Fotios K. Konstantinidis
Institute of Communication and Computer Systems (ICCS), National Technical University of Athens, Heroon Polytechniou 9, Zografou Campus, 15773 Athens, Greece
3 shared publications George Arvanitakis
3 shared publications- Angelos Amditis
National Technical University Of Athens Institute of Communications and Computer Systems, Zografou, Attica, 15773, Greece
2 shared publications
Latest Publications
- Open Access
- 18 June 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (6) , 267
A Deep Learning Approach for Pixel-Level Material Classification via Hyperspectral Imaging
Recent advancements in computer vision, particularly in detection, segmentation, and classification, have significantly impacted various domains. However, these advancements are still strongly tied to RGB-based systems, which are insufficient for applications in industries such as waste sorting, pharmaceuticals, and defence, where material characterization beyond shape or visible colour is necessary. Hyperspectral (HS) imaging captures spatial and spectral information for each pixel and therefore offers a promising route for material-level classification. This study evaluates the potential of combining HS imaging with deep learning for plastic material classification. The work includes: (i) the design of an experimental setup with a HS line-scan camera, conveyor, and controlled illumination; (ii) the construction of an object-disjoint dataset of HDPE, PET, PP, and PS samples with semi-automated mask generation and Raman spectroscopy-based labelling; and (iii) the development of P1CH, a lightweight pixel-wise 1D convolutional hyperspectral classifier. On object-disjoint test images, P1CH achieved 97.44% all-pixel accuracy. A boundary sensitivity analysis, reported separately because semi-automated labels are uncertain at material/background interfaces, yielded 99.94% accuracy after excluding a pre-defined two-pixel border band. Additional ablation, baseline, and robustness analyses show that the proposed pixel-wise spectral approach is effective for small fragments, visually similar plastics, and overlapping materials, while black or very dark plastics remain challenging under the present camera and illumination configuration.
- 18 June 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (6) , 267
Open AccessA Deep Learning Approach for Pixel-Level Material Classification via Hyperspectral Imaging
Recent advancements in computer vision, particularly in detection, segmentation, and classification, have significantly impacted various domains. However, these advancements are still strongly tied to RGB-based systems, which are insufficient for applications in industries such as waste sorting, pharmaceuticals, and defence, where material characterization beyond shape or visible colour is necessary. Hyperspectral (HS) imaging captures spatial and spectral information for each pixel and therefore offers a promising route for material-level classification. This study evaluates the potential of combining HS imaging with deep learning for plastic material classification. The work includes: (i) the design of an experimental setup with a HS line-scan camera, conveyor, and controlled illumination; (ii) the construction of an object-disjoint dataset of HDPE, PET, PP, and PS samples with semi-automated mask generation and Raman spectroscopy-based labelling; and (iii) the development of P1CH, a lightweight pixel-wise 1D convolutional hyperspectral classifier. On object-disjoint test images, P1CH achieved 97.44% all-pixel accuracy. A boundary sensitivity analysis, reported separately because semi-automated labels are uncertain at material/background interfaces, yielded 99.94% accuracy after excluding a pre-defined two-pixel border band. Additional ablation, baseline, and robustness analyses show that the proposed pixel-wise spectral approach is effective for small fragments, visually similar plastics, and overlapping materials, while black or very dark plastics remain challenging under the present camera and illumination configuration.
- Open Access
A Deep Learning Approach for Pixel-level Material Classification via Hyperspectral Imaging
Recent advancements in computer vision, particularly in detection, segmentation, and classification, have significantly impacted various domains. However, these advancements are tied to RGB-based systems, which are insufficient for applications in industries like waste sorting, pharmaceuticals, and defense, where advanced object characterization beyond shape or color is necessary. Hyperspectral (HS) imaging, capturing both spectral and spatial information, addresses these limitations and offers advantages over conventional technologies such as X-ray fluorescence and Raman spectroscopy, particularly in terms of speed, cost, and safety. This study evaluates the potential of combining HS imaging with deep learning for material characterization. The research involves: i) designing an experimental setup with HS camera, conveyor, and controlled lighting; ii) generating a multi-object dataset of various plastics (HDPE, PET, PP, PS) with semi-automated mask generation and Raman spectroscopy-based labeling; and iii) developing a deep learning model trained on HS images for pixel-level material classification. The model achieved 99.94\% classification accuracy, demonstrating robustness in color, size, and shape invariance, and effectively handling material overlap. Limitations, such as challenges with black objects, are also discussed. Extending computer vision beyond RGB to HS imaging proves feasible, overcoming major limitations of traditional methods and showing strong potential for future applications.
Open AccessA Deep Learning Approach for Pixel-level Material Classification via Hyperspectral Imaging
Recent advancements in computer vision, particularly in detection, segmentation, and classification, have significantly impacted various domains. However, these advancements are tied to RGB-based systems, which are insufficient for applications in industries like waste sorting, pharmaceuticals, and defense, where advanced object characterization beyond shape or color is necessary. Hyperspectral (HS) imaging, capturing both spectral and spatial information, addresses these limitations and offers advantages over conventional technologies such as X-ray fluorescence and Raman spectroscopy, particularly in terms of speed, cost, and safety. This study evaluates the potential of combining HS imaging with deep learning for material characterization. The research involves: i) designing an experimental setup with HS camera, conveyor, and controlled lighting; ii) generating a multi-object dataset of various plastics (HDPE, PET, PP, PS) with semi-automated mask generation and Raman spectroscopy-based labeling; and iii) developing a deep learning model trained on HS images for pixel-level material classification. The model achieved 99.94\% classification accuracy, demonstrating robustness in color, size, and shape invariance, and effectively handling material overlap. Limitations, such as challenges with black objects, are also discussed. Extending computer vision beyond RGB to HS imaging proves feasible, overcoming major limitations of traditional methods and showing strong potential for future applications.
- 24 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 359-363
Ore Sorter as Part of Mining 4.0 from Mechatronics Approach
While the Industry 4.0 principles are taking over more industrial sectors, artificial intelligence and cutting-edge sensors and actuators are constantly developed. However, this technological rise causes higher requirements in raw material extractions. To achieve this demand, the mining industry is trying to adopt the guidelines of Mine 4.0 elevating their procedures of extraction, control and monitoring, while also disengage the human intervention from certain mining operations, such as the sorting procedure of extracted minerals. To accomplish this scope, this research enters the Cyber-Physical Sorting System (CPSS) in mining industry, which provides adaptable features for implementation in various value chains. In particular, the CPSS described in this paper will be applied in the mining field, enhancing the magnesite sorting process using RGB and hyperspectral imaging inline and X-Ray Fluorescence (XRF) analysis and RAMAN spectroscopy offline ensembles with deep convolutional neural networks for classification and an automated air nozzle system for the final separation.
- 24 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 359-363
Ore Sorter as Part of Mining 4.0 from Mechatronics Approach
While the Industry 4.0 principles are taking over more industrial sectors, artificial intelligence and cutting-edge sensors and actuators are constantly developed. However, this technological rise causes higher requirements in raw material extractions. To achieve this demand, the mining industry is trying to adopt the guidelines of Mine 4.0 elevating their procedures of extraction, control and monitoring, while also disengage the human intervention from certain mining operations, such as the sorting procedure of extracted minerals. To accomplish this scope, this research enters the Cyber-Physical Sorting System (CPSS) in mining industry, which provides adaptable features for implementation in various value chains. In particular, the CPSS described in this paper will be applied in the mining field, enhancing the magnesite sorting process using RGB and hyperspectral imaging inline and X-Ray Fluorescence (XRF) analysis and RAMAN spectroscopy offline ensembles with deep convolutional neural networks for classification and an automated air nozzle system for the final separation.
- 17 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Exploration and Mitigation of the Impact of Lighting Conditions on Multi-spectral Image Classification
Imaging spectroscopy is a potent and versatile technique, offering valuable spatial and spectral data across a diverse range of applications such as agriculture, environmental monitoring, material characterization and remote sensing. Nevertheless, due to its passive nature, multi-spectral and hyperspectral sensors heavily rely on appropriate illumination sources, especially in scenarios where natural sunlight might be insufficient or unavailable. To comprehend the challenges associated with illumination source selection, an extended experimental effort has been launched to investigate the effects of different lighting conditions on multi-spectral image analysis. A dataset of images captured from both the visible (16 bands) and Near-Infrared (25 bands) spectral regions under Full-Spectrum halogen and visible-spectrum LED illumination sources, was generated. Within this dataset, four distinct types of plastic waste, namely High Density PolyEthylene (HDPE), Low Density PolyEthylene (LDPE), PolyEthylene Terephthalate (PET), as well as PolyPropylene (PP), are incorporated. A set of machine learning models aim to distinguish the different kind of plastics. The models’ accuracy, was employed as metric to quantify the impact different lighting conditions have on multi-spectral image classification. This work underlines the necessity of appropriate illumination sources on such imaging applications, while providing crucial insights to mitigate the effects of light conditions on the classification performance.
- 17 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Exploration and Mitigation of the Impact of Lighting Conditions on Multi-spectral Image Classification
Imaging spectroscopy is a potent and versatile technique, offering valuable spatial and spectral data across a diverse range of applications such as agriculture, environmental monitoring, material characterization and remote sensing. Nevertheless, due to its passive nature, multi-spectral and hyperspectral sensors heavily rely on appropriate illumination sources, especially in scenarios where natural sunlight might be insufficient or unavailable. To comprehend the challenges associated with illumination source selection, an extended experimental effort has been launched to investigate the effects of different lighting conditions on multi-spectral image analysis. A dataset of images captured from both the visible (16 bands) and Near-Infrared (25 bands) spectral regions under Full-Spectrum halogen and visible-spectrum LED illumination sources, was generated. Within this dataset, four distinct types of plastic waste, namely High Density PolyEthylene (HDPE), Low Density PolyEthylene (LDPE), PolyEthylene Terephthalate (PET), as well as PolyPropylene (PP), are incorporated. A set of machine learning models aim to distinguish the different kind of plastics. The models’ accuracy, was employed as metric to quantify the impact different lighting conditions have on multi-spectral image classification. This work underlines the necessity of appropriate illumination sources on such imaging applications, while providing crucial insights to mitigate the effects of light conditions on the classification performance.
- Open Access
- 11 October 2023
- Published by Elsevier in Resources, Conservation and Recycling
- Vol. 199
Multi-modal sorting in plastic and wood waste streams
- 11 October 2023
- Published by Elsevier in Resources, Conservation and Recycling
- Vol. 199
Open AccessMulti-modal sorting in plastic and wood waste streams
- Open Access
- 13 January 2023
- Published by Elsevier in Procedia Computer Science
- Vol. 217, 227-237
Multi-sensor cyber-physical sorting system (CPSS) based on Industry 4.0 principles: A multi-functional approach
- 13 January 2023
- Published by Elsevier in Procedia Computer Science
- Vol. 217, 227-237
Open AccessMulti-sensor cyber-physical sorting system (CPSS) based on Industry 4.0 principles: A multi-functional approach