ce399 | research archive: (anti)fascism

Facial Profiling (SF Weekly 14-20/7/10)

Posted in Uncategorized by ce399 on 23/07/2010

Facial Profiling
Will face-recognition technology get an accused killer off the hook?
By Peter Jamison
published: July 14, 2010

Courtesy Ben Bavarian
Courtesy of Ben Bavarian
Courtesy of SFPD
Prosecutors said Richard Barrett was killed for this distinctive pendant depicting Bamm-Bamm from The Flintstones.
Prosecutors said Richard Barrett was killed for this distinctive pendant depicting Bamm-Bamm from The Flintstones.
Eartha Goodwin
Barrett’s stepfather and mother outside the courthouse on the day Heard’s verdict was announced.
Barrett’s stepfather and mother outside the courthouse on the day Heard’s verdict was announced.
Frank Gaglione
Eric Safire, Heard’s defense lawyer, said surveillance videos showed his client could not be the shooter.
Eric Safire, Heard’s defense lawyer, said surveillance videos showed his client could not be the shooter.
Courtesy of Laura Barrett
Barrett was shot twice in the back after leaving a North Beach strip club in 2008.
Barrett was shot twice in the back after leaving a North Beach strip club in 2008.

Peter Jamison on face-recognition technology
The science of biometrics is exactly what it sounds like — measurement of the living body, including quantification of such physical characteristics as fingerprints, eyes, and voice. But in recent years it has come to be associated most immediately by laypersons with one particular type of biometric analysis: facial recognition.

In the years after the terrorist attacks of Sept. 11, 2001, techniques for precisely identifying strangers by their most public and easily recorded attributes — their faces — seemed to hold great promise as a means for improving national security. If terrorists at airports, shopping malls, or government buildings could be easily picked out when their images appeared on surveillance cameras, then other ruses for bypassing security safeguards, such as phony passports or birth certificates, could be foiled.

That was the promise. The reality has consisted largely of disappointment. While the U.S. Department of Homeland Security and the Department of Defense remain bullish on the future of facial biometrics — the latter has poured hundreds of millions of dollars into research and development of face-recognition techniques — the technology has yet to live up to its early hype during the post–9/11 period.

Police departments in Florida and Virginia have abandoned facial-biometrics surveillance systems that did not yield any arrests while they were deployed. Several years ago, a surveillance experiment at a train station in Mainz, Germany, found that automated facial recognition had a success rate of only 60 percent during the day and as low as 10 percent at night, when poor lighting made identification more difficult. The American Civil Liberties Union has also raised questions about the government’s research into facial recognition, arguing that the technology violates privacy and often produces mismatches.

As a result, facial-identification biometrics has failed to gain widespread acceptance in the law-enforcement and scientific communities for most uses. But last month in San Francisco, it was abruptly adopted for a task far more specific — and arguably more serious — than those for which it has been considered in the past: the identification of a killer.

In what legal and scientific experts say is a groundbreaking case, a San Francisco Superior Court judge allowed biometric facial-identification technology, along with accompanying testimony from an expert witness, to be admitted as evidence in a high-profile criminal trial.

And in a curious turn, biometrics — the science popularized for its use in attempts to catch terrorists — was being used in San Francisco to try to exonerate an accused gang member and murderer.

The case was that of Charles “Cheese” Heard, 25, an alleged Western Addition gangster accused of murdering San Francisco resident Richard Barrett during an attempted robbery in North Beach on Nov. 25, 2008.

Business surveillance cameras captured footage of the man who probably shot Barrett, as well as an accomplice. Heard’s defense team persuaded a judge to allow testimony from an expert who argued, based on a biometric comparison of still frames from the video and photos taken of Heard while he was in jail, that the defendant could not be the same person as the shooter.

This was an unusual development for a couple of reasons. It’s uncommon for criminal-defense attorneys to seek to have new scientific evidence admitted to court; far more often, it is prosecutors who try to foist novel techniques for identifying suspects on jurors, as viewers of TV’s Forensic Files can attest. In the case of biometrics, the irony was particularly striking, since most of the technology’s previous applications have been in settings where it was used to screen and apprehend suspects, not to set them free.

More significantly, Heard’s case could serve as a precedent — albeit not a legally binding one, since it was not ruled on by the state’s appellate or supreme court — for the future use of facial-identification technology in criminal trials. In a motion opposing the defense’s efforts to present its biometrics analysis to the jury, Assistant District Attorney Michael Swart noted that no California appeals court has ever ruled on whether this form of technology could be used as evidence.

David Faigman, an expert on scientific evidence at Hastings College of the Law in San Francisco, said the judge’s decision to allow facial-recognition techniques into the courtroom was a surprise, since the technology appears not to meet a few basic legal standards applied to other forms of scientific evidence, such as DNA or fingerprint analysis.

Faigman said that given how little we know about the reliability of facial-identification biometrics, the admission of such evidence in Heard’s trial could be seen as a troubling development, particularly if other judges accept the case as a precedent for future trials.

“I think it is precedent-setting,” he said in an interview. “But I also think that the appellate courts might take a dim view of the admission of this evidence. … Without the systematic and rigorous evaluation of the evidence, it’s hard to know how much weight to give it.”

Nonetheless, biometrics was set to play a starring role in front of an audience of San Francisco jurors in the Heard trial. Would they deem the technology good enough to help set an accused murderer free?

Richard Barrett was 29 years old when he was killed shortly after leaving Larry Flynt’s Hustler Club, a well-known gentleman’s club off a notoriously vice-ridden and violent strip of Broadway in North Beach. According to cops and prosecutors, Barrett was shot twice in the back by thieves trying to steal one of his prized possessions: a jewel-encrusted pendant depicting the cave-baby Bamm-Bamm from The Flintstones.

“Bam bam” is street slang for cocaine, and during Heard’s trial, his defense attorney, Eric Safire, argued that Barrett, while the victim of a tragic crime, was a criminal himself. According to Safire, Barrett sold drugs and mingled with the characters — strippers, whores, pimps, gangbangers, and the boozy bridge-and-tunnel crowd — who walk Broadway’s streets after dark. “He was selling cocaine,” Safire told the jury. “It could have been a bad drug deal … somebody could be mad he’s pimping their girlfriend.”

Prosecutors and police had a different story. They argued that Heard was a member of the Central Divis Playas gang in the Western Addition, and that he specialized in jewelry robberies. Barrett, they said, was Heard’s latest victim.

The trial was an odd one from the start. The attorneys presented a distinct contrast in styles. Safire is short and bald, with a neatly trimmed goatee. He sports fancy ties and matching kerchiefs inside the courtroom, and a Panama hat outside.

His adversary, Swart, is tall and shambling. He has blond hair and boyish features that are often set in a scowl of irritation. Despite his stature, he resembles a sullen teenager shoved into a lawyer’s suit.

In October, in what turned into a carnival-esque preliminary hearing, the D.A.’s office flew its star witness to San Francisco from Texas to identify Heard in the courtroom. Francis Smith had been in San Francisco in November 2008 on a business trip, and was partying with co-workers in North Beach when she saw Barrett shot.

Frustrated that his client was not allowed to take part in a line-up, Safire arranged for seven young men — all, like Heard, black and wearing gold “grills” across their front teeth — to stand in the back of the courtroom and stare at the witness at the precise moment she was asked to identify the shooter. Smith was not fazed by the gambit, and calmly picked out Heard where he sat at the defense table.

Swart was another story. He sprung out of his chair, shouting that “out-and-out intimidation of a witness” had taken place, and ordered the seven gold-toothed agitators arrested. They were cuffed and led away in a scrum of police officers and sheriff’s deputies, the corridors of the Hall of Justice echoing with shouted curses while Safire looked on, shaking his head.

“It was a circus,” Tina Heard, Charles’ mother, recalled of her son’s months in the courtroom.

A part-time cosmetologist who lives in Tracy and commutes to work on the graveyard shift at the Guittard Chocolate Company in Burlingame, she believes her son was having the book thrown at him because some SFPD officers had mistakenly decided that he was a hardened gangster. “They’re so determined to put my son away that they will accept anything, and go to any extremes,” she said in an interview.

As it turned out, the prosecution couldn’t find too many extremes to go to. Aside from Smith’s testimony — the linchpin of Swart’s case — much of the evidence against Heard was circumstantial.

Cellphone records showed that Heard might have made several calls from the neighborhood where Barrett was shot soon after the killing, although one of the calls was to his mother, arguably not a gang member’s likeliest confidant in the aftermath of a murder. An FBI wiretap caught Heard bragging about his habit of robbing people’s bling (“When they put on some jewelry, they no longer my niggas,” he explained in one recording), but no specific mention of the Barrett robbery.

The case had other weaknesses. Heard never actually stole the Bamm-Bamm pendant: The necklace was removed from Barrett’s body at the scene of the crime by an unconnected bystander and returned to the victim’s family. Swart argued that the killing had been an act of bravado to prevent the alleged gang member from “losing face” with the rest of his Central Divis Playas after Barrett resisted the theft.

The defense also had its own eyewitnesses — two of them — who claimed to have seen the shooter and said it wasn’t Heard. One of them, David Stribble, a co-worker of Smith’s, gave particularly compelling testimony. When asked whether Heard was the man he had seen shoot Barrett, he replied, “Not even close.”

The trial of Charles Heard was, in other words, what criminal lawyers call an “identity case.” The defendant’s guilt or innocence rested on jurors’ willingness to believe various statements about what the killer of Richard Barrett looked like.

It was into this setting that there walked a beguiling, self-described expert who claimed he could settle the question using a little-known branch of science.

From the get-go, Ben Bavarian made no secret of his eagerness to participate in a San Francisco criminal proceeding that, outside of its potential ripeness for his profession’s techniques, had absolutely nothing to do with him.

“I think this is going to be the first time that we use the fundamentals of digital facial attributes in a court case,” he said in an interview during his first trip to San Francisco last month for a hearing on whether his testimony should be admitted in Heard’s murder trial. “It’s going to be a very important case — a historical case.”

Bavarian is managing director of Newport Beach–based firm AFIS and Biometrics Consulting, a private consultancy that develops biometric identification systems. He holds a Ph.D. in electrical and computer engineering from Ohio State University, and — with his amber complexion, coiffed dark hair, and hooded gaze — exhibits a distinct Southern California gloss. His clients have included the Department of Homeland Security and the Netherlands national police.

Safire had enlisted Bavarian to analyze business surveillance videos of two young black men fleeing the site of Barrett’s murder. The prosecution and defense agreed that one of these unidentified men was the shooter. They wore clothes matching those described by witnesses, and one can be seen in the footage holding an object that resembles a handgun.

Swart chose not to use the videos for identification purposes, asserting that they were too blurry. Yet they were a mainstay of Safire’s case. Relying on Bavarian’s techniques, the defense lawyer hoped to demonstrate that neither of the men captured on tape was Charles Heard.

The prosecution, for its part, argued forcefully against Bavarian being allowed any-where near a jury.

“At this point, it cannot be established that biometrics has been generally accepted in the relevant scientific community,” Swart said in a motion. “Just as importantly, there are no prior published Court of Appeal opinions in California that establish the general scientific acceptance of this science.”

Swart’s characterization of facial-identification technology is accurate, according to Faigman: “My experience suggests that it’s not generally accepted in the mainstream scientific community,” he said.

Bavarian defended his proposed use of biometrics during an interview with SF Weekly, claiming that facial identification could do a lot more than the prosecutor was willing to admit. “This is not a junk science,” he said. “This is a good science.”

The defense lucked out: Judge Jerome Benson decided that Bavarian could take his turn in the witness box. “I’m not satisfied that the procedures he used are improper,” Benson said after both attorneys had argued the question of admitting Bavarian’s testimony. “The motion to admit the doctor’s testimony is granted, with substantial limitations.”

Benson ruled that Bavarian could not explicitly state, based on his analysis, that the man in the surveillance video and Charles Heard were different people. However, he said, “He can say the measurements in the video surveillance footage are whatever they are, and the measurements in the jail photo [of Heard] are whatever they are,” and let the jury ponder the difference.

Despite such restrictions, jurors would almost certainly be able to infer the gist of Bavarian’s testimony — that Heard was not the man in the video, and hence was not guilty.

When Bavarian took the stand on June 10, he brandished a laser pointer that he directed at a slideshow on the wall opposite the jury. The courtroom was dark, and the jurors leaned forward in their seats.

Two sets of images recurred throughout the presentation. The first was a straight-on shot of Heard in jail-issue orange, staring blankly at the camera. The second was an eerie black-and-white still frame of the supposed murderer caught on surveillance tape, displaying an angular face twisted into a smile and shadowed by a hoodie.

Bavarian explained that his method was to take measurements connecting various points on each individual’s pictured face, turn them into ratios (for instance, creating a ratio of the distance between a person’s eyes to the distance between his hairline and chin), and then match them against each other. By this method, he said, the biometric signatures of individual faces — even in photos of different size and resolution — could be contrasted.

“These are the numbers you came up with,” Safire said to his expert. “Do they show a difference?”

Bavarian stated that the difference was, indeed, “significant in the context of other measurements.”

Swart’s time to cross-examine the doctor came. “Sir, your background is a background in electrical engineering. Is that true?” the prosecutor asked.

“No,” Bavarian said. “Electrical and computer engineering.”

“Have you gone through a course in forensic photography?”

Bavarian tried to ask a question.

“Let me ask the question, and then you can answer the question,” Swart said.

He started peppering Bavarian with queries designed to undermine his credibility.

“Do you belong to the Scientific Working Group on Image Technology?”

“I just became a member of that,” Bavarian replied.


“Last month.”

“Have you been admitted to the Facial Identification Scientific Working Group?”

“Yes,” Bavarian said. Then, after a pause: “That’s not relevant.”

“It’s not relevant? Well, I’m glad you can make rulings for us on what’s relevant and what’s not,” Swart said. “Thank you.”

Safire, seated at the defense table, grinned.

After some interrogation on the technical aspects of Bavarian’s methods, Swart asked him whether his analysis of Heard’s photo and that from the surveillance camera had been peer-reviewed by other scientists.

“Did anybody do a technical review of your work?”

“Yes,” Bavarian replied.


“My colleagues and my friends.”

Safire objected.

“Your Honor, I think this is misleading,” he said somberly. “I really do.”

If it seems odd that a relatively untested science, explained by a relatively unknown scientist, should be considered as evidence by a panel of men and women deciding whether to send a young man to prison for the rest of his life — well, it is.

State and federal courts ostensibly have rigorous tests for establishing the admissibility of scientific evidence. The goal, as a California appellate judge wrote in 1998, is to avoid bestowing “a misleading aura of certainty or a posture of mystic infallibility” on quackery.

“The traditional fear is that the jury will much more readily accept something that an expert says is true,” said Michael Saks, a professor at Arizona State University College of Law and an expert on scientific evidence. “The official doctrine is, you want to move slowly and carefully and not let something in until it’s really good.”

Yet this doctrine has been more honored in the breach than in the observance, according to Saks. Many novel brands of forensic science have been heedlessly allowed into high-stakes trials, and withdrawn from courtroom use only when later called into question. Forms of “scientific” evidence that have been accepted and then debunked include voice-print identification, bullet-lead analysis, blood-spatter studies, handwriting, bite marks, and certain burn patterns that were once thought to indicate arson.

Strange to say, in the 19th century, before forensic fingerprinting became established, a primitive form of biometrics that measured people’s heads and limbs — known as Bertillonage after its creator, a French police clerk named Alphonse Bertillon — was the dominant means of identifying suspects in the criminal-justice system.

“We have a long and somewhat ambivalent tradition of using biometrics for legal identity,” said Jennifer Mnookin, a professor at UCLA School of Law and an expert on forensics and scientific evidence. “The question is, how well does it work? And how well does it work from a blurry photograph?”

There’s a simple answer to that question, according to some biometrics authorities: not well. Whatever Bavarian’s friends may have thought about his work on the Charles Heard case, it turns out that some of his colleagues have reservations about it.

Academic experts in facial-recognition techniques interviewed by SF Weekly, along with an FBI forensics specialist who testified on behalf of the prosecution in Heard’s trial, expressed skepticism both about facial recognition’s readiness for the courtroom and the specific methods used by Bavarian.

“Any biometrics system can make a mistake,” said Anil Jain, a Michigan State University professor of computer science and engineering and a biometrics expert. “In the case of face, the accuracy is worse, because face changes with respect to illumination, with respect to aging. … I think face recognition, particularly for surveillance applications, is not quite ready for automated identification.”

Jain also raised questions about Bavarian’s comparison of the images of Heard and the video of the supposed shooter. After a phone interview with SF Weekly, he reviewed the “Biometric Analysis” submitted to the court by Bavarian and, in a subsequent e-mail, highlighted what he saw as several weaknesses in the report.

In Heard’s case, Jain noted, “It is not surprising that they have an inconclusive result given the quality of video. … The problem of comparing a high-resolution image with the low-resolution video is challenging.” Additionally, he noted that Bavarian’s decision to perform a comparison of the faces in the images by hand, rather than using a computer program, “introduces a lot of subjectivity.”

Asked about these criticisms, Bavarian said he had performed the analysis manually because the low-quality images from the surveillance tapes could not be accurately fed into automated biometrics software. He acknowledged a risk of subjectivity in his method, but said it had been minimized — and the poor quality of the images corrected for — by comparing Heard’s jail photo to dozens of still frames of the shooter gleaned from the video footage, rather than just one or two.

Perhaps the most damning review of Bavarian’s work came from the expert witness Swart put on the stand to debunk his analysis.

Richard Vorder Bruegge is a forensic scientist and photographic technologist employed by the FBI. He has a Ph.D. in geological sciences from Brown University, a full head of silver hair, and youthful good looks, presenting a stark contrast to Bavarian’s vaguely reptilian allure.

The authoritativeness of Vorder Bruegge’s résumé was almost comical. He worked with NASA on the Clementine mission, which carried out the first image mapping of the moon in 1994; he is the chairman of the Facial Identification Scientific Working Group — the same group Bavarian had claimed was “not relevant”; and he came to San Francisco for Heard’s trial just two weeks after delivering the keynote address on facial identification at a biometrics conference in Australia.

On the stand, Vorder Bruegge argued that the technique used by Bavarian — measuring the distances between facial features, creating ratios, and then comparing them — was “not reliable,” because it depended too much on such variables as a subject’s pose and the angle of the camera. A more fundamental problem in Heard’s case, he said, was that the image gleaned from the low-resolution surveillance video simply wasn’t fit for scientific analysis.

“Dr. Vorder Bruegge,” Swart asked, as he concluded his questioning, “in your opinion, is Dr. Bavarian’s report reliable?”

“No,” Vorder Bruegge said.

Swart repeated, “Is it reliable?”

“No,” Vorder Bruegge said evenly. “It is not.”

As the trial wound down, Safire chose to highlight the blurry surveillance video as one of several factors pointing to Heard’s innocence. The biometric evidence, he asserted, was simply one more factor that should raise a reasonable doubt as to whether Heard had committed murder.

The prosecutors in the case, he told jurors in his closing argument, were “trying to scare you into making a mistake. Why? Because Mr. Heard’s image is not in that video.” Of the biometric analysis, he said, “Listen, you guys heard the scientific evidence. I don’t quite get it myself, but all I can tell you is that was an attempt to try to make this case easier to decide.”

During his own summation, Swart set to bashing Bavarian again. He also praised the testimony of Vorder Bruegge.

“A guy who used to work for NASA came in here, and he said [the video is] just too low-resolution,” he said.

“He exposed Dr. Bavarian for what he is,” he continued. “In the legal community, we have a term for what Dr. Bavarian tried to do. It’s called junk science.”

Charles Heard’s jury deliberated for nine days. It was an improbably long period for a relatively uncomplicated murder case. Earlier this year, a San Francisco jury returned not-guilty verdicts in a five-month double-murder trial after a little more than a day.

As deliberations stretched on, it became clear that jurors were struggling with the issue that had been at the heart of the case from the beginning — the identity of Barrett’s killer. The jury eventually asked the judge for guidance on the “felony murder doctrine,” under which defendants can be convicted of a murder they did not personally commit, so long as they were participating in the crime that led to the killing.

The meaning of this communication was clear to those versed in criminal law. Despite the testimony of multiple eyewitnesses and the biometric analysis provided by Bavarian and countered by Vorder Bruegge, the jury couldn’t settle the question of whether Heard had been the shooter.

Ultimately, jurors decided he had not been. But they still convicted him of first-degree murder.

On the morning of July 1, the jury announced that it had found Heard guilty under the felony murder rule — though he could not be identified as the shooter, the jurors believed he had participated in the attempted robbery that resulted in Barrett’s death — and guilty of attempted robbery. It hung on a third charge, for firearm possession.

When the verdict was announced, a cheer went up from Barrett’s family members, who had packed the gallery. Outside the court, his relatives were effusive, crying as they voiced their joy at the case’s outcome.

“I’m really happy. Justice was served for my family, my granddaughters,” said Laura Barrett, the victim’s mother. “For them to shoot him in the back like that — Charles Heard ain’t nothin’ but a coward.”

Nobody expressed any doubt as to whether prosecutors had charged the right man. To the contrary: Barrett’s relatives said the streets had been abuzz with rumors that Heard had been the killer.

Exactly one week after finding him guilty of first-degree murder, Heard’s jury hung on allegations that he was a gang member. A separate deliberation for these charges took place; if convicted, he could have faced a stiffer minimum prison sentence.

Nevertheless, the case was a clear victory for the D.A.’s office. Even without the gang charges, Heard faces a sentence of 26 years to life.

“The jury convicted a person that is a very violent person, and the community is a lot safer because of their actions,” Swart told SF Weekly.

Safire said he would appeal the guilty verdicts. He plans to argue that the case’s outcome was flawed because the jury both rejected the prosecution’s theory that Heard was the shooter and found him guilty of murder.

“The defense prevailed, and lost anyway,” he said. “You can’t find as a jury that he shot the guy and didn’t shoot the guy.”

In the end, did the biometric evidence matter? Did the jury turn to the felony murder doctrine because Bavarian’s techniques planted a seed of doubt as to whether Heard was the hooded man in the grainy videotape?

One juror who spoke to SF Weekly following the verdicts said the opposite was true. The juror, who did not want his name published, said that based on his own scientific background — a computer programmer by profession, he also majored in mathematics in college — he didn’t find Bavarian credible, and had given more weight to Vorder Bruegge’s countertestimony.

“We had two experts,” he said. “One was believable and one wasn’t. … It was really a reach for what [Bavarian] was trying to do.”

Without speaking to the other jurors, it’s impossible to know how the biometric analysis may have influenced them. But it seems possible that Heard’s case has opened the door to a controversial form of scientific evidence that could prove more consequential in the future, whether used to set people free or put them away.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: