H2I - Hyperspectral Images for Inspection Applications

An EFRE-FESR funded project (FESR1111)

Responsive image


The Hyperspectral Images for Inspection Applications (H2I) project is an EFRE/FESR funded project (FESR1111) coordinated by the Faculty of Computer Science of the Free University of Bozen-Bolzano and Microtec.

The H2I project aims at creating an hyperspectral image acquisition platform and a set of Deep Learning algorithms able to deal with hyperspectral images. Scroll down to find out more!


  • We present the H2I project at the Industry Day of the IEEE World Congress on Computational Intelligence (IEEE WCCI 2022) in Padova on July 20, 2022.
  • We talk about 'Hyper-spectral image classification for wood recognition' at the SFScon2021 that will take place on November 11-12, 2021. Check out the presentation and the video here.
  • We are busy organising a H2I workshop that will take place on September 16, 2021. Check the program here.
  • Our paper 'A Lightweight Approach for Wood Hyperspectral Images Classification' was accepted at the IEEE International Conference on Multimedia and Expo (ICME 2021) Industry Track!


The precise knowledge of some phenomena and physical details of organic and non-organic materials has been the subject of extensive scientific research for years. Some of these features, such as defects in wood or fruit, are not easily detected with the technologies currently available. To detect and analyse these phenomena, there is a need for both reliable and fast technology to collect images of materials, and software solutions capable of generating useful information for humans, possibly in a very precise way and in real-time, to meet the needs of the industrial world.

Hyperspectral Images allow the composition of objects in a scene to be examined non-destructively. This is achieved by generating the band spectrum of each pixel in the scene. Hyperspectral Images are widely used in remote sensing applications, astronomy, and, more recenlty, in archaeology, agriculture and in the wood and food industry.

There are two main problems related with Hyperspectral Images. First, hardware platforms are expensive and complex. Second, extracting useful information from Hyperspectral Images is complicated because the spectral dimension is present in addition to the spatial dimension.


The main goals of the H2I project are:

  • To develop a Hyperspectral Images acquisition platform with a configurable hyperspectral sensor, and a dedicated light source with a tunable spectral illumination profile.
  • To develop a set of Deep Learning-based techniques to deal with spatial and spectral information, specifically suited for hyperspectral data.

The acquisition platform and the Deep Learning methods could be used in a variety of applications, such as inspecting wood recognition, wood's fungi, checking the ripeness of fruits and identifying any internal diseases that may be present, or checking the condition of archaeological artefacts.


The dissemination of scientific results is an essential part of research. To this end, we will organise a workshop on 'Hyperspectral Images for Inspection Applications: The Use-case of Wood and Fungi Detection'.

The workshop will last for one day and will take place on September 16, 2021. Participation in the workshop will be free via a Zoom meeting, upon a registration through the following google form: register .

The workshop will be scheduled as follows:

  • Welcome and Introduction (Roberto Confalonieri – UniBZ, Tammam Tillo – IIIT-Dehli)
    • 9:00-9:15: Workshop program overview (slides)
  • Data modelling using Hyper-spectral images (Matteo Caffini - Microtec, Simone Faccini - Microtec, Philipp Bock - Microtec)
    • 9:15-10:00: Wood industry
    • 10:00-10:30: Hyper-spectral images and images acquisition (slides)
  • 11:00-11:15 - Break
  • Data classification using Deep Learning (Roberto Confalonieri – UniBZ, Tammam Tillo – IIIT-Dehli)
    • 11:15-12:15: Deep Learning for Images Classification and Hyperspectral Images Classification Framework (slides)
  • 12:15-13:45 - Break
  • Hands-on sessions (Phyu Phyu Htun– University of Computer Studies Yangon, Boyuan Sun – UniBZ, Attaullah Buriro – UniBZ, Salim Malek – FBK, Davide Cremonini - UniBZ)
    • 13:45-14:15: Spatial Classification of Hyper-spectral images (slides)
    • 14:15-14:45: Spectral Classification of Hyper-spectral images (slides)
    • 14:45-15:30: Data Augmentation of Hyper-spectral images (slides_Part_I, slides_Part_II)
    • 15:30-16:00: Fungi detection (slides)
  • 16:00-16:15 - Break
  • Wood analysis using X-ray and molecular techniques (Roberto Terzano – UniBA, Carlo Porfido – UniBA, Youry Pii - UniBZ)
    • 16:15-16:45: Wood analysis using tomographic and X-ray microfluorescence techniques (slides)
    • 16:45-17:00: Towards molecular analysis of wood
  • 17:00-17:15 - Closing and Wrap-up

The program of the workshop can be downloaded here.

People Involved

  • Roberto Confalonieri, Lead Partner PI (UniBZ)
  • Tammam Tillo, Technical Coordinator (IIIT-Dehli)
  • Youry Pii, Associate Professor (UniBZ)
  • Matteo Caffini, Research Engineer (Microtec)
  • Simone Faccini, Research Engineer (Microtec)
  • Marco Boschetti, Technology scouting (Microtec)
  • Philipp Bock, Head of Product Development (Microtec)
  • Attaullah Buriro, Researcher (UniBZ)
  • Boyuan Sun, Researcher (UniBZ)
  • Phyu Phyu Htun, PhD student (University of Computer Studies, Yangon, Myanmar)
  • Davide Cremonini, Master Student (UniBZ)
  • Salim Malek, Researcher (Fondazione Bruno Kessler, and previously UniBZ)

Contact us

Please contact us for more information.