2017 - GENETIC ALGORITHMS FOR OPTIMIZATION OF LIGHTING DESING

Lighting design for interior spaces is the process of integrating artificial light sources into architectural complexes. This work proposes a novel approach to interior lighting design, leveraging genetic algorithms and 3D computer graphics. The system exploits the software Blender to reproduce the architectural space and simulate the effect of illumination. Secondly, the system implements the genetic algorithm NSGA-II to solve the multi-objective optimization problem of inverse illumination.

2016 - SIMULATING FIRE HAZARD IN INDUSTRIAL ENVIRONMENT

This work presents a novel system to simulate the risk of fire outbreaks in complex industrial plants. The system uses the software Blender to render the high-quality model of a real oil refinery. Then, the system simulates the dynamics of the flames and predicts how the heat affects critical components of the industrial plant. The proposed system is integrated into a wider safety system for risk assessment and emergency management.

2015 - RECOMMENDING ATTRACTIVE ROUTES WITH FLICKR

This work proposes a recommendation system of urban routes that are not necessarily the shortest (or fastest), but the most attractive. The system leverages VGI (volunteered geographic information), geographic data extracted from images and videos uploaded on Flickr. A clustering algorithm finds the locations with the highest density of popular geo-tagged posts. Then, the system converts these locations into intermediate waypoints for the optimal route.

2015 - MOTIF DISCOVERY FOR SEMANTIC CLASSIFICATION OF NATURAL LANGUAGE

This work proposes to recognize semantic characteristics of natural language by extracting recurring motifs. The system exploits the tool FRED to transform text into formal structured graphs using RDF/OWL ontologies. The system analyzes the resulting network to discover recurring motifs, which could indicate functional relations in the sentence.

2014 - IMAGE QUILTING ALGORITHM FOR TEXTURE SYNTHESIS

This work proposes an image quilting algorithm to synthesize large-scale seamless textures. The algorithm uses a patch-based approach, which generates a new texture (of arbitrary size) by stitching together small patches of an existing source image. The patches are stitched together along the path of minimum cost, so that the generated texture is seamless.