Empower will pilot test a total of 8 investigative tools in the fields of:
IMAGE/VIDEO
This tool consists of a hardware device for autonomous video surveillance, installable in vehicles and powered by battery. The device will be able to receive images from several cameras at the same time.
This tool can be used to collect and preserve digital evidence and support or refute hypotheses before courts. In particular, the person re-identification tool is inspired on the image retrieval concept.
The video summarisation tool is a forensic application that can be used to reach a fast triage, data deduplication and advanced data enhancement of large amount of data from privileged data sources.
This tool detects logos in unconstrained content. The tool will decouple the feature extraction from the detection functionality, taking advantage of self-supervised techniques that lead to better knowledge modelling.
TEXT/VOICE
Calligraphus is an automatic transcript generation and edition platform based on proprietary speech processing technology (AUDIMUS).
Transkit is a technological asset supported in Deep Neural Networks for transcription of multimedia content, turning audio into text for later post-editing and exploitation.
The Clustering tool organises large volumes of multi-modal data in coherent groups with each focusing on a specific topic whilst being distinct from the rest of identified groups.
FEDERATED LEARNING
This framework will assist LEA partners to cooperate in fully leveraging the potential of their private and decentralized data.
in a concerted effort carried out by a consortium of 10 partners from 5 Member States (Greece, Ireland, Italy, Portugal and Spain), representatives of the Security and Law Enforcement R&I community: 2 LEAs (MIR-PN and PJ), 6 companies and SMEs (SLG, HERTA, ICTLC,TGO, TREBE, and VOI) and 2 RTOs (CERTH and VICOM).
As a result of the project, 8 tools will be brought to an operational deployment level with testing by MIR-PN and PJ with operational datasets in real-life environments. Such tools will be made available to other non-partner LEAs through open repositories and licensing schemes in coordination with key players such as Europol, EACTDA, ENFSI, EuCB and CERIS. Lastly, anonymised data sets, a training needs analysis, training materials and recommendations about i) Data Interoperability and standardisation, ii) AI Trustworthiness of LEA tools, and iii) the Innovation Uptake models of new AI based solutions by LEAs, will also be widely disseminated.
CONSORTIUM
SISTER PROJECT
A big-data analytics from base-stations registrations and CDRS e-evidence system.
