AI for Media and Humanities

The Artificial Intelligence for Media and Humanities laboratory has the mission to investigate and advance the state of the art in the Artificial Intelligence field, specifically addressing applications to digital media and digital humanities, and taking also into account issues related to scalability.

Multimedia

Investigating new AI-based solutions to image and video content analysis, understanding, and classification. This includes techniques for detection, recognition (object, pedestrian, face, etc), classification, feature extraction (low- and high-level, relational, cross-media, etc), anomaly detection also considering adversarial machine learning threats.

Human Language

Investigating AI-based solutions to textual data analysis, understanding, and classification. This includes representation learning for text classification, transfer learning for cross-lingual and cross-domain text classification, sentiment classification, sequence learning for information extraction, text quantification, transductive text classification, and applications of the above to domains such as authorship analysis and technology-assisted review.

Humanities

Investigating AI-based solutions to represent, access, archive, and manage tangible and intangible cultural heritage data. This includes solutions based on ontologies, with a special focus on narratives, and solutions based on multimedia content analysis, recognition, and retrieval.

Large-scale IR

Investigating efficient, effective, and scalable AI-based solutions for searching multimedia content in large datasets of non-annotated data. This includes techniques for multimedia content extraction and representation, scalable access methods for similarity search, multimedia database management.

Highlight

In this annual report, we sum up the activities of the AIMH research group on 2020. We summarize the research conducted on our main research fields, describe the projects in which we were involved, report the complete list of papers we published, together with their abstract, list thesis on which we were involved, highlight the datasets we created and made publicly available.

Highlight

5G-Enabled Objects and Persons recognition for Unmanned Aircraft

Public safety has become a relevant aspect, especially in large urban areas. The possibility to act promptly in case of dangers or alarms can be of fundamental importance in determining the favorable outcome of the interventions. For example, identifying and tracking an individual or a suspicious vehicle that moves in an urban context…