Liguria Regional Operational Program 2014 - 2020
Co-financed by F.E.S.R. - European Regional Development Fund
Axis 1 "Research and Innovation (OT1)"
Action 1.2.4
Call - "Support for the implementation of complex research and development projects for companies associated with research and innovation poles"
CUP: G32C21000910007
The MoreThanThis™ Project (MObility REgulation THinking ANd THoroughly Innovative inSpired") is a RD initiative coordinated by HITACHI RAIL™ addressed to innovation in the Smart Stations sector.
As part of the project, knowhedge™ has developed an advanced video surveillance analysis solution based on the use of AI algorithms and neural networks for the automatic detection of anomalies and defects caused by the presence of dust, condensation, drops of water, breakage of the lens/camera casing, occlusions, poor pointing and/or lighting conditions, etc. . The project has been tested and validated at high TRL during video sequences acquired by the surveillance cameras installed in the De Ferrari station of the Genoa Metro (North West Italy).
The research objective was to provide a cutting-edge infrastructure based on the AIoT platform of the subsidiary MYWAI SRL that combines new Generative AI and Edge AI algorithms by generating synthetic data on the cloud to train algorithms on Edge computing.
The developed technology allows to support the maintenance activities of the video surveillance system through automation
identification, reporting and certified management of data relating to operating anomalies.
Given the limited availability of real data representative of the selected anomalies, necessary for the training and verification of the recognition models, as part of the research knowhedge™ has developed and fine-tuned data augmentation algorithms and tested Generative AI methods for the synthesis of anomalous images of various types starting from real images acquired by the cameras being tested.