Retrospective Facial Recognition System

The MPS will benefit from an updated Retrospective Facial Recognition (RFR) search capability to enable a more effective use of images and image frames from video data across all types of investigations. Technical advancements made over recent years would, if seized, now allow the MPS opportunities that were not previously available to support the detection and matching of faces. The opportunity also represents a chance to realise significant savings in terms officer time it takes to reconcile an image of a person to that person’s identity. This helps prevent and detect crime and keeps Londoners safe.

The Deputy Mayor for Policing and Crime is recommended to approve this request to go to the commercial market via CDW, the MPS approved Software Reseller, for a Retrospective Facial Recognition system. This will allow the MPS to identify a suitable product and vendor and is a compliant route to market.

1. Introduction and background

1.1. The MPS will benefit from a Retrospective Facial Recognition (RFR) search capability to enable the effective use of images and image frames from video data across all types of investigations. In parallel it will enable the MPS to effectively exploit those investigative opportunities that have occurred with the sharp growth in these image data sources. Rapid and ongoing advancements in this technology would, if seized, now allow the MPS opportunities that were not previously available to support the detection and matching of faces even when processing lower quality images and videos.

2. Issues for consideration

2.1. The main purpose of RFR searching is to assist in identifying suspects from still images or specific images extracted from video. These images will need to be lawfully held by the MPS. These may be images that have been captured by cameras at burglaries, assaults, shootings and other crime scenes. They could also be images shared by or submitted by members of the public. As well as assisting in preventing and detecting crime, RFR searching could also be used to help in the identification of missing or deceased persons.

2.2. The RFR use case is very different to Live Facial Recognition and seeks to help officers identify persons from media of events that have already happened and does not involve members of the public walking past the system ‘live time’. As such it would be a tool that helps aid the investigative process, by analysing still images or images that have been specifically extracted from a media source. The result of this analysis will present investigators with additional leads to consider.

2.3. Human-in-the-loop decision is a critical aspect of the RFR proposal and will be embedded into RFR processes.

3. Financial Comments

3.1. This proposal does not seeking funding at this point.

4. Legal Comments

4.1. Any use of facial recognition technology needs to be in accordance with the law. The legal framework includes the Data Protection Act 2018, Human Rights Act 1998, Equality Act 2010 and The Protection of Freedoms Act 2012. The process of identifying a suitable vendor will enable diligence to be undertaken as regards matters such as statistical accuracy and demographic performance so that legal compliance can be determined.

4.2. The Mayor’s Office for Policing and Crime (“MOPAC”) is a contracting authority as defined in the Public Contracts Regulations 2015 (“the Regulations”). All awards of, and modifications to, public contracts for goods and/or services valued at £189,330 or above shall be procured in accordance with the Regulations.

4.3. Paragraph 4.13 of the MOPAC Scheme of Delegation and Consent provides the Deputy Mayor for Policing and Crime (DMPC) has delegated authority to approve the procurement strategy of and requests to go out to tender for contracts valued at £500,000 or above.

5. Commercial Issues

5.1. This is a request to approve a compliant route to market via the CDW value added reseller framework. This reflects a need to undertake significant vendor diligence prior to taking any proposal forwards.

5.2. An RFP (request for proposal exercise) will be conducted utilising the Framework with the intention to award to a single supplier who can offer competitive pricing and compliance with the MPS specification of requirements.

5.3. As part of the RPF and during the assessment of products the MPS shall:

• take reasonable steps to satisfy itself that the product selected does not have an unacceptable demographic differential such that the MPS can be satisfied of complying with its PSED duties and data protection obligations;

• assess the statistical accuracy and demographic performance of the product including comparisons between products considered for use.

6. GDPR and Data Privacy

6.1. The MPS is subject to the requirements and conditions placed on it as a 'State' body to comply with the European Convention of Human Rights and the Data Protection Act (DPA) 2018. Both legislative requirements place an obligation on the MPS to process personal data fairly and lawfully in order to safeguard the rights and freedoms of individuals.

6.2. Under Article 35 of the General Data Protection Regulation (GDPR) and Section 57 of the DPA 2018, Data Protection Impact Assessments (DPIA) become mandatory for organisations with technologies and processes that are likely to result in a high risk to the rights of the data subjects.

6.3. The Information Assurance and Information Rights units within MPS will be consulted at all stages to ensure the project meets its compliance requirements.

6.4. A DPIA will be required for this project and the selection of the vendor is crucial to the completion of the DPIA. This is because the DPIA and the associated control measures needed to mitigate any risk identified will be informed by the statistical accuracy demographic performance of the selected vendor’s product. The project will ensure a privacy by design approach, which will allow the MPS to find and fix problems at the early stages of any project, ensuring compliance with data protection legislation. DPIAs support the accountability principle, as they will ensure the MPS complies with its legal duties.

7. Equality Comments

7.1. The use of facial recognition technology engages the Public Sector Equality Duty (PSED). Amongst other points, the PSED means that there is a need to take reasonable steps to understand the performance of a facial recognition algorithm. The process of identifying a vendor at this point would enable the MPS to assess its ability to comply with the PSED when proposing to take any vendor forwards.

8. Background/supporting papers

8.1. Report.