The FBI said it has received a “tremendous” amount of video from the public taken near the Marathon’s finish line where bombings killed three people and stunned the nation Monday.
Crucial evidence could turn up in video shot on smartphones and other recorders. The challenge: How to sort through all those images quickly and effectively.
Experts say analyzing so much content is a mountainous task and believe it will take months to process the digital evidence from what Boston police commissioner Edward F. Davis called the most photographed event in the country that day.
“It’s a very tedious process,” said Andrew Obuchowski Jr., an associate director at Navigant Consulting Inc. in New York and a former law enforcement forensic video analyst. “Trying to be able to get through it quickly is a challenge.”
Advertisement
The process is cumbersome due to the limits of software that automatically processes such content. Investigators still must watch most of it and then enter information into a database by hand.
Law enforcement officials in Canada took this approach to 5,000 hours of video evidence captured in the 2011 Vancouver hockey riots. A special team of forensic experts spent two weeks manually processing the data, which captured thousands of crimes.
Experts say this part of the Boston investigation could take four to five times as long, based on the sheer volume of amateur imagery captured.
“Every single person down at the Marathon who had a cellphone was collecting evidence for police,” said Grant Fredericks, who led the team that analyzed the video evidence in Vancouver. “They didn’t know it at the time, but they probably have every inch of that event covered for every second of that day.”
Investigations into the attacks on London’s public transit system in 2005 and the Times Square bombing attempt in 2010 were also aided by video and photo evidence.
Advertisement
Authorities in Boston have urged spectators to submit video and photos taken near the finish line during the race and terrorist attack that injured 176 people. Police have also retrieved surveillance content from surrounding businesses.
Police can use video analytic software to sort though content from fixed-position cameras, searching for movement in a specific area or color. That tool can track such content because it is shot in a consistent, unmoving frame of view. But there is no software capable of tracking and analyzing cellphone content in the same manner, so investigators will have to review those images frame by frame.
In most cases, facial recognition software can’t be applied to cellphone videos, either.
“When we start talking about mobile phone and hand-held cameras, the video is unstable,” said Larry Compton,of Forensic Video Solutions Inc., a company he leads with Fredericks. “It’s shaking all over the place, the camera is moving, and you don’t have any fixed point of reference.”
In the Vancouver case, police called in the Law Enforcement and Emergency Services Video Association, a nonprofit professional organization deployed to conduct forensic video analysis in terrorist attacks.
Analysts used a program called Omnivore to capture the videos and then a second program, Avid Media Composer, to view and tag sections. Tagging uses software codes to log information about people on screen, times, actions, and other information police can search for later.
“By adding these data points, investigators in the field can later determine they need to identify a person in a black-hooded sweatshirt in a certain area in a specific time frame and then go back and search the database,” Obuchowski said.
Advertisement
Fredericks said the time it takes to analyze these videos often hinges on the amount of data about the recordings available to investigators. Cellphone users, for example, must enable their device to track geographical information. If evidence is submitted without that data or police don’t ask for it, the puzzle becomes more difficult to piece together, Fredericks said.
Police in Vancouver stumbled when they didn’t specifically ask for raw, unedited videos. Fredericks said that some videos couldn’t be authenticated and used in court because the recordings had been edited before they were submitted to police. In other cases, e-mailed videos had to be compressed into appropriate file sizes, which reduced the quality.
Thus far, the FBI has advised anyone with images or videos to e-mail them to boston@ic.fbi.gov or to call 800-CALL-FBI.
FBI Special Agent Richard DesLauriers said yesterday that a specialist from the FBI lab in Quantico, Va., was going to help analyze videos from before, during, and after the blasts.
Fredericks said members of his association, which operates from a lab at the University of Indianapolis, have been in touch with the FBI. He said they had not yet been asked to assist the investigation.
Taryn Luna can be reached at taryn.luna@globe.com.