Notice

Real-World Incident Response Hackathon Information Sheet

Published 9 November 2018

1. General

For this event, we are looking to bring together the best from academia, industry and government in the defence and security arena.

Participants will be asked to process large amounts of real-world incident multimedia data and rapidly identify key information for on-site experts - people, places, events, in fact anything that may assist an investigations team. We’re interested in how entities relate to each other, the event timeline and narrative, and near-term predictions. Data will mostly take the form of video which will be varied in quality, source, and format. Some will be live-streamed during the event. The majority will be sourced from CCTV cameras.

This is a unique opportunity for participants to demonstrate their ability to extract useful information and insights from large multimedia data sources which would help teams respond to incidents more quickly and effectively. We expect participants to exploit cutting-edge artificial iIntelligence techniques, including machine vision, facial recognition, and object recognition to achieve the best results.

2. Background

When a major incident occurs, the police and security services are inundated with data from both private and public sources. Sorting through this data is challenging as it may be gathered out of order, from multiple locations, and is of varying quality and format.

This data is primarily video, with a significant amount sourced from CCTV cameras, but also includes still images, text reports, and social media posts. Not all the data shared with the investigations team is relevant. Video data is presented in multiple formats, resolutions, bit rates, frame rates, angles, colour spaces (including monochrome and black and white), and codecs. This may include data in proprietary formats, such as those used by CCTV systems. A variable amount and accuracy of metadata is associated with each file.

A single incident will generate several terabytes of data for analysis. Some of the data that requires analysing is live-streamed.

The response team analyse this data to identify the key people, places, events, and objects relating to the incident. This is a time-consuming process and is often done under extreme time pressure. The team needs to rapidly determine what happened, who caused it, where that person came from, and where they went.

3. Aim of the hackathon

This hackathon is looking to improve the way the investigations team responds to incidents. Several terabytes of historical real-life data, relating to a single incident, will be provided with teams asked to extract as much actionable data from the data set as possible.

The investigators need to have relevant content surfaced to them quickly from within a large data set. We expect participants to utilise Artificial Intelligence techniques to identify people, places, and objects of interest in footage. As the hackathon will utilise real-world data, it is highly representative of the sorts of challenges faced by the investigations team. The aim of the hackathon is to determine whether there are tools available, or in development, that can work within these constraints.

At the end of the two-day hackathon, participants will be asked to present their solutions to a panel of expert judges.

4. Future State

When an incident occurs, the investigations team can take in a large volume of disparate, multi-format, multi-media data and begin analysing it. From a single starting point (for example, a single video of an incident occurring) they can quickly and accurately identify the relevant people, objects, and events. Of primary importance is the identification of potential suspects, and tracking their appearance across multiple camera feeds and still images.

5. Benefits of Attending

Attendees of the event will network and collaborate with likeminded experts in their field creating teams to tackle a challenge that has potential commercial applications across a range of business sectors.

It is your chance to:

• develop and test your digital product to boost its potential
• showcase your prototype to international defence and science professionals and technology experts
• seize the opportunity to take your solution to Minimum Viable Product (MVP)

The team outputs will be assessed, and prizes awarded to the winners.

The first prize is a guided tour through the underground Burlington Bunker in Corsham, Wiltshire. This is a unique opportunity to visit a complete underground city, built in the 1950s to house the UK’s entire cabinet in the event of a nuclear attack. Sitting 1,000 feet underneath Corsham, the bunker was built to house 4,000 people in comfort for up to three months and includes over 60 miles of roads, a hospital, a reservoir, canteens, a bakery, a television studio, and what was at the time Britain’s second largest telephone exchange. The existence of Burlington Bunker was classified until 2004. This prize is subject to confirmation and suitability of the winners.

We are looking for solutions that:

• deliver a working proof of principle to improve the speed and accuracy of processing data
• will work with some or all of our legacy systems
• meet security requirements
• are evergreen and cyber resilient
• are cost effective
• can demonstrate significant benefit
• can be easily accessed by multiple users through a common user interface
• have potential to be scaled up in the future

We are not looking for:

• theoretical solutions or IS architecture
• consultancy
• paper-based studies or literature reviews
• solutions that do not offer significant benefits
• PhD proposals
• projects that can’t demonstrate feasibility within the timescale
• solutions that do not work with existing systems

6. Who Should Attend

We’re extending this invitation to all interested parties including companies with existing solutions which may be tailored to this problem during the event, individuals who can offer some unique expertise, and everyone in between. Individuals will be asked to join others in teams, with facilitated matchmaking taking place at the start of the event.

This is a unique opportunity to work with real data and spend quality time with real end users. It is a technical event and as such we are expecting participants to be of a technical nature. This is not a sales event, and as such non-technical sales staff should not attend.

7. Exploitation

No contract will be awarded as a result of this event. If a solution of this nature is purchased in the future, standard commercial processes will be followed.

8. Development Environment

Further information will be in the guidance document that will be issued to registered participants.

A development environment will be provided by Microsoft at no cost to participants:

• data will be pre-loaded on to Azure storage prior to the event. - this can be accessed either directly through Azure (storage can be mounted in an Azure hosted Virtual Machine, for example) or through standard Azure APIs and endpoints which are accessible outside of Azure, including from alternative cloud providers like Amazon Web Services.
• a detailed list of tools and resources being made available by Microsoft will be shared before the event in the guidance document.
• a Microsoft representative will attend the event as a support / mentor / guide / coach for those teams who want to use Azure. On-line sessions will be available prior to the event to answer any questions participants may have regarding Azure.

Use of this environment is optional. Participants are welcome to use their own hardware, and software, including cloud based solutions. If a participant wishes to bring their own hardware (larger than a standard desktop, or with unique power / space requirements), please inform the organisers at least 5 working days before the event so that appropriate provisions can be made.

At several points during the event, video footage will be live-streamed to participants. To process these feeds, participants will need to be able to connect to an internet end point.

The venue will provide free, unlimited WiFi during the event with no filtering or captive portal.

9. User behaviour and approach

CCTV and multimedia are important strands of any police investigation. CCTV can be the first aspect of evidence in ascertaining what, when, and how an incident occurred. CCTV is exported onto removable media by police trawl/recovery officers from the immediate incident location such as from shops, offices, private premises or occasionally remotely accessed where surveillance systems are connected to police control rooms. In addition, imagery sourced from members of the public, often in the form of mobile phone recordings are also used to form a picture of events prior to and following an incident. This imagery comes in a multitude of media formats that need to be independently viewed by analysts, using time and location as their primary indicators in order to identify subjects/objects of interest. Positive subject/object of interest identification can then lead to analysts tracking them or it across various CCTV or multimedia sources both prior to and following the incident with a view to obtaining the best possible image for circulation. This viewing and tracking task often directs further CCTV collections as and when updated subject/object location information is obtained.

10. Range of users and roles

The team comprises of 12 police trawl/recovery officers who are responsible for source CCTV identification and timeframe collection, 12 police analysts who view imagery searching for what, when and how, directing further CCTV recovery where necessary, and 3 technical engineers dealing with problematic recovery or viewing issues.

11. Business Rules

Participants would benefit from understanding CCTV Guidelines and general digital evidence handling guidance for UK police services.

12. Additional Information

Information regarding the format or content of data will not be provided in advance of the event. This is to understand the flexibility of participants’ ability to deal with previously unseen data and accurately reflects the challenges faced by the investigations team.

Following registration closing on Friday 16 November, registered participants will receive the following:

• hackathon guidance pack
• terms and conditions
• non-disclosure agreement (which must be signed prior to the event)