Projects

The application of Artificial Intelligence in environmental monitoring can greatly impact a better and safer quality of life. It can influence the food, water, and air quality, with clear improvements in real life.

Starting from this assumption, any application of AI in these specific areas requires special attention and dedication.

The early non-invasive detection of plant diseases is a challenging area with a relevant impact on environmental preservation. The goal of our research in this field is to define the spectral signatures of target diseases (Xylella fastidiosa, Chrysochroa fulgidissima) that are devasting vegetation in the South of Italy. Hyperspectral images acquired by a Specim AFX10 sensor mounted on a UAV (DJI Matrice 600 Pro) are processed by anomaly detection tools, to distinguish infected trees from healthy ones before the effects of the disease become visible.*

*This research activity is supported by a research project titled ENDOR - ENabling technologies for Defence and mOnitoring of the foRests - Research Program PON DFM.AD001.347 of the Italian MIUR

In the last years, the marine habitat has been under pollution threat; this impacts many human activities as well as human life. Increasing concerns about pollution levels in the oceans and coastal regions have led to multiple approaches for measuring marine pollution, in order to achieve sustainable marine water quality.

The focus of this research activity is to detect the presence of marine pollution by means of multispectral and hyperspectral imaging systems. The sensors have been mounted on a UAV (DJI Matrice 400), and recorded images have been processed my using CNN in order to detect (multispectral analysis) and classify (hyperspectral analysis) macroplastics.*

*This research activity is supported by a research project titled SIRIMAP - Detection Systems of marine plastic pollution and subsequent recovery-recycling - Research Program PON ARS01_01183 of the Italian MIUR

Vehicle re-identification is currently one of the most important topics within the scientific community.

The focus of this project is the vehicle re-identification based on the Convolutional Neural Network (CNN) to implement a methodology that can learn the salient characteristics of a vehicle, and exploits them for the re-identification of the same in different areas, or at different times.

Experiments have been performed on VeRI 776 datasets*.

*Special thanks to Xinchen Liu, the administrator of the VeRi dataset for vehicle re-identification

The goal of this project is the automatic monitoring of Marine Protected Area finalized to:

- people detection and counting;

- intrusion detection;

- erosion monitoring;

- boat detection and tracking

This project focuses on a case study for automatic visual monitoring of welding process, in dry stainless steel kegs for food storage. 

In the considered manufacturing process the upper and lower skirts are welded to the vessel by means of Tungsten Inert Gas (TIG) welding.

During the process several problems can arise: 1) residuals on the bottom 2) darker weld 3) excessive/poor penetration and 4) outgrowths. The proposed system deals with all the four aforementioned problems and its inspection performances have been evaluated by using a large set of kegs demonstrating both the reliability in terms of defect detection and the suitability to be introduced in the manufacturing system in terms of computational costs.