SOO - F - Tracking Criminal Actors - 06022021.pdf

Tracking Criminal Actors in Large Crowds

June 2021 Science and Technology Directorate Department of Homeland Security

DISTRIBUTION NOTICE: Further dissemination only as directed by DHS Office of Policy DESTRUCTION NOTICE: Destroy by any method that will prevent disclosure of contents or reconstruction of the document.

1.0 Problem Statement

Ability to identify and mark or "tag" active/known criminals in large crowds discreetly or overtly for future apprehension

2.0 Purpose and Scope

When conducting crowd control activities, law enforcement officers and first responders do not have a solution that enables them to track (overtly or covertly) active or known criminals, or individuals within large groups who engage in criminal activities. First responders are currently limited by the human capital required to monitor crowds and lack a full motion technology solution that can identify and track criminal activities. The solution would need to work in real time and in a variety of conditions, including low light or poor visibility. It is essential that the solution be reliable and able to differentiate between the intended criminal target versus other persons who may be dressed the same, wearing disguises, or otherwise appear similar. Of utmost importance, the solution must uphold civil liberties, existing laws and privacy applications.

3.0 Background Environment

Law enforcement lacks the ability to accurately and consistently track criminal actors in large crowds, particularly criminals that may be using the cover of large crowds or lawful protests to camouflage their criminal actions and evade detection, identification and arrest. Currently, law enforcement relies on undercover officers or CCTV. Undercover operations are inconsistent and may be hazardous to the officers. CCTV has a host of issues from privacy to reliability and challenges with identification of perpetrators with masks and other attempts to conceal their identity.

In the Spring and Summer of 2020, first responders across the country were responsible for providing crowd control during protests and demonstrations that marked widespread civil unrest. In some circumstances individuals within the crowds committed criminal offenses. The tools and methods that first responders must identify and track criminal actors in large crowds is insufficient.

This SOO is intended to address the need for a new or enhanced capability that will enable first responders to perform their required missions, enforce the law, and ensure their safety and the safety of others.

3.1 Mission Scenarios

Civil unrest occurred in small towns and large cities across the United States in 2020, requiring the nation’s first responders to manage substantial crowds of protesters and demonstrators that at times became violent. Some events were attended by known criminals, while others involved persons within the crowd committing spontaneous criminal acts. Law enforcement needs the ability to identify known criminal as well as offenders that are unknown at the time of the commission of their crime. Law enforcement needs the ability to track and identify criminals so that they may be detained at a time of tactical advantage to law enforcement.

3.2 Current Technology Shortfalls and Threats

Human capital is one of the biggest shortfalls for first responders when it comes to tracking criminal actors in large crowds. First responders are limited in their ability to track multiple persons at once.

They can’t leave their assigned post to track a person(s) as he or she moves through a crowd or follow an individual until they are “arrestable.” Tracking criminal actors requires one observer for each target, which doesn’t work in large crowds and is very difficult to do in situations where there are numerous criminal actors in one location.

In addition to gaps pertaining to human capital, the FRRG identified the following shortfalls with respect to existing technology used to track criminal actors in large crowds:

• There is no system able to tag and track people covertly and in real time

• Unable to quantify behaviors through technology without getting a lot of false positives

• Unable to differentiate between the intended target and someone who may be dressed similarly or look similar

• Facial recognition software is insufficiently advanced to be reliable and safe, does not work well in crowds, and has issues with recognizing black and brown faces

• Unable to do gait and pixel tracking in real time (a capability that exists if reviewing the video historically)

• Unable to “pre-tag” an individual using composite images or videos to train the system to look for and track specific persons

• Technology isn’t interoperable with social media, to include using social media images or videos for tagging or tracking

• Communication between the technology (or persons using the technology) is insufficient, relying upon group text messages

• Videos aren’t currently co-mingled, resulting in them having to be pieced together after the event occurs

• There are gaps in cost and training for existing technologies

4.0 Potential Solution Attributes

First responders require a solution that can address the full spectrum of crowd control scenarios they are responsible for managing. The solution will allow for covert and/or overt tracking of criminal actors and may be achieved through either a mechanical technology or a software solution using multiple feeds of video or a combination of technologies. This SOO does not seek to restrict the path to solution and is primarily concerned with the delivered capability.

4.1 Concept of Operations (ConOps)

The solution will provide first responders with a technology that is able to do the following:

• Draws a polygon or identify a subject overtly/covertly throughout the video or system

• Provides pictures or tracking information that is shareable with field units

• Is taggable with criminal predicate characteristics

• Inputs or pre-programs known criminal targets to allow for tagging of persons of interest

• Can accept input from any fixed or streaming video service

• Has the ability to scrape output data from other existing systems while maintaining the resolution of the parent system

• Can visualize the pathways of the person(s) identified, see their movements

• Can record the information and can rewind the data in real time in order to review it

• Provides an end user experience that:

Includes a user-friendly GUI Is presented in a digestible format Enables information to be layered on and off as needed Is integrated with geospatial standard technology, to include providing an overhead view and a camera view Is portable or wearable, such as a mobile application or wearable device that provides information and alerts including an auditory proximity alert that indicates when a tagged person is nearby

The FRRG members indicated the solution could leverage existing COTS or GOTS technologies, or parts of existing technologies, that may be currently be in use. This includes, but is not limited to, technologies used by the Department of Defense or in the United Kingdom to conduct surveillance or contract tracing. They also indicated there may be technologies used to tag and track persons during sporting events, for example those used by the National Football League or NASCAR, that could also be informative.

4.2 Additional Capabilities

Successful development of this requirement will enable first responders to be better prepared to help safeguard their communities, to include those lawfully practicing their First Amendment rights.

Potential future applications of the solution identified by the FRRG could include providing the capability to:

• Extrapolate information to identify cars or other objects of interest

• Recognize and track firearms or other weapons or devices

• Allow for identification of specific logos or branding being worn by persons within the crowd

5.0 Solution Capabilities and Requirements

The FRRG members identified must-have requirements for the deployment and design of the solution, as well as desirable capabilities or characteristics, which are outlined in the sections below.

5.1 Must Have Requirements

The must have characteristics of the solution the FRRG identified include:

• Identifies and tracks criminal activities

• Upholds existing laws, civil liberties, and privacy applications

• Meets all relevant standards

• Accepts inputs from any fixed or streaming video service

• Considers the full range of potential data sources, including aerial or drone footage, web-based streaming, social media

• Scrapes data from existing outputs from other systems while maintaining like resolution of the parent system

• Ensures sources of video or inputs are acceptable to be used as evidence in a court of law

• Retains and stores data in the system, ownership of all system data belongs to the user

• Uses an intuitive GUI, like the iPhone

• Provides readily available training, help files, tutorials, and information (“just in time training”)

• Includes built in help features (mouse hover, flyouts)

• Pushes data (images, etc.) out to responders in the field in real time, for example via a mobile application

• Ensures file transfers are secure, encrypted end-to-end and while at rest

• Provides the option to take notes, annotate

• Ensures interoperability and is compatible with existing systems

• Uses a relatively universal (nonproprietary) file type

• Provides an application programming interface (API) allows for open source data or other data sources to feed into it

• Includes a percentage or degree of confidence for the likelihood of each potential match

• Allows automatic and manual targeting for known faces or criminal behaviors; enables operator to direct system to target individual

• Stitches multiple video feeds together to provide situational awareness, three-dimensional modeling

• Differentiates between the intended target and other individuals who may be dressed the same (at the time of identification), wearing disguises, or otherwise appear similar

• Continues to track an individual who has been identified even if they change clothing

• Allows for geographic boundaries (e.g., up to ten miles) or physical locations to be set

• Provides adequate protections to ensure the system cannot be hacked or leveraged for use against first responders.

5.2 Desirable Capabilities or Characteristics

In addition to the must have requirements reported in subsection 5.1, there are performance and/or system standards that exceed the proposed solution requirements:

• Covert tagging and tracking ability

• Ability to have data uploaded from the field into the command post

• Ability for system to have only historic inputs (e.g., Boston bombing white/black hat tracking)

• Provides option to scan or observe in real time without recording, then start recording when behavior is observed and continue recording for a given period

6.0 System Support

6.1 Contractual Requirements

6.1.1 Period and Place of Performance

The length of time of the contract shall be between 12-18 months and shall not exceed the latter. All developmental activities will be performed at the contractor-designated site(s). DHS may require the contractor to perform critical design reviews and/or product presentations at DHS-specified locations.

6.1.2 System Affordability

First responders’ budgets are limited and should be taken into consideration when developing and determining a price per unit cost model. Given the variation in regional assets and requirements, the solution should be scalable and provide add ons and features that are optional so it can remain affordable.

6.1.3 Regulatory/Legal Requirements

The solution must me all applicable standards and regulations. This includes, but is not limited to, the following:

• Federal encryption standard, FIPS compliant, 140-2

• Records retention timeframes

• Privacy standards, PIA In addition, any source of video or input will need to be acceptable as evidence in a court of law.

With the lack of a national standard for file or document type recognized industry standards should be utilized in any solution.

6.1.4 Contract Deliverables

The project performer will be required to provide several prototypes by the end of the period of performance for post-contract testing and evaluation. The number of prototypes will vary depending on cost and complexity, however the FRRG expects the need for two to five.

6.2 Solution Support

6.2.1 Support Concept

6.2.2 Training

Training information as required shall be provided on any specialized equipment, and shall include how to properly store, clean, deploy, and maintain the basic ensemble/unit. Any specialty requirements of personnel conducting maintenance, storage, extraction, or delivery will be identified including certification requirements. Training material shall be available initially in hard-copy, and subsequently in soft-copy format.

6.2.3 Other Considerations

Human Interface Requirements: The target user is law enforcement, but solutions may also be applicable to EMS. The solution should be able to integrate with other equipment/PPE currently used by these emergency responders. It should also be able to operate in all types of environments and in temperature extremes.

Logistics and Readiness: The solution developed under this award is envisioned as an enhanced version of already available technology. It is envisioned that the solution will be portable and integrated as to increase responder efficiency which ultimately saves more lives.

6.2.4 Storage and Maintenance

The contractor shall identify all necessary storage and maintenance requirements for the proposed solution including: the required maintenance schedule, identification of the specific components of the ensemble that requires the maintenance, the type of maintenance required, and whether the maintenance can be performed in the field by operational personnel or if it will be required to be returned to the manufacturer for the required maintenance.

Physical solutions including PPE, shall include all special laundering instructions, including recommendations for frequency of laundering, should also be provided if required. Laundering recommendations should be viable and cost effective for First Responders and First Responder agencies and should ideally be able to be performed “in house.”