`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`_____________________
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`_____________________
`
`
`GLOBAL TEL*LINK CORPORATION
`Petitioner
`
`v.
`
`SECURUS TECHNOLOGIES, INC.
`Patent Owner
`____________
`
`Case IPR2016-01220
`Patent 9,007,420 B1
`__________________
`
`
`PETITIONER GLOBAL TEL*LINK CORPORATION’S REPLY TO PATENT
`OWNER RESPONSE
`
`
`
`
`
`
`
`Mail Stop “PATENT BOARD”
`Patent Trial and Appeal Board
`U.S. Patent and Trademark Office
`P.O. Box 1450
`Alexandria, VA 22313-1450
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`PETITIONER’S UPDATED EXHIBIT LIST
`
`
`Description
`
`U.S. Patent No. 9,007,420 to Passe, filed January 10, 2014, issued
`April 14, 2015.
`Prosecution History of U.S. Patent No. 9,007,420.
`U.S. Patent Application Publication No. 2012/0262271 by
`Torgersrud et al., filed April 18, 2011, published October 18, 2012.
`U.S. Patent No. 8,218,829 to Kenoyer, filed July 31, 2006, issued
`July 10, 2012.
`U.S. Patent No. 7,436,988 to Zhang et al., filed June 3, 2005, issued
`Oct. 14, 2008.
`Declaration of Dr. Hoyamoon Beigi.
`E. Hjelmås et al., “Face detection: A Survey,” Computer Vision and
`Image Understanding, Vol. 83, No. 3, September 2001; pp. 236-74.
`M. Yang et al., “Detecting Faces in Images: A Survey,” IEEE
`Transactions on Pattern Analysis and Machine Intelligence, Vol. 24,
`No. 1, January 2002; pp. 34-58.
`C. Zhang et al., “A Survey of Recent Advances in Face Detection,”
`Microsoft Research Technical Report No. MSR-TR-2010-66,
`Microsoft Corporation, 2010.
`M. Kirby et al., “Application of the Karhunen-Loeve Procedure for
`the Characterization of Human Faces,”, IEEE Transactions on
`Pattern Analysis and Machine Intelligence, Vol. 12, No. 1, January
`1990; pp. 103-08.
`M. Turk et al., “Eigenfaces for Recognition,” Journal of Cognitive
`Neuroscience, Vol. 3, No. 1, 1991; pp. 71-86.
`J. Cartoux et al., “Face Authentication or Recognition by Profile
`Extraction from Range Images,” IEEE Proceedings of the
`Workshop on Interpretation of 3D Scenes, November 1989; pp.
`194-99.
`K. Bowyer et al., “A survey of approaches and challenges in 3D and
`multi-modal 3D+2D face recognition,” Computer Vision and Image
`Understanding, Vol. 101, No. 1, 2006; pp. 1-15.
`
`GTL
`Exhibit No.
`1001
`
`1002
`1003
`
`1004
`
`1005
`
`1006
`1007
`
`1008
`
`1009
`
`1010
`
`1011
`
`1012
`
`1013
`
`
`
`- i -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`
`Description
`
`U.S. Patent No. 9,064,257 to Beigi, filed November 2, 2011, issued
`June 23, 2015.
`U.S. Patent No. 6,219,640 to Basu et al., filed August 6, 1999,
`issued April 17, 2001.
`P. Viola et al., “Rapid Object Detection Using a Boosted Cascade of
`Simple Features,” Computer Vision and Pattern Recognition, 2001.
`L.Sanchez Lopez, “Local Binary Patterns Applied to Face Detection
`and Face Recognition,” Final Research Project, Universitat
`Politecnica de Catalunya, November 2010.
`Y. H. Huang et al., “Face Detection and Smile Detection,”
`Proceedings of IPPR Conference on Computer Vision, Graphics,
`and Image Processing, Shitou, Taiwan, A5-6, 2009.
`M. Felsberg, “Robot Vision Systems, PhD course spring term
`2015,” Computer Vision Laboratory PowerPoint Presentation, 2015.
`INTENTIONALLY LEFT BLANK
`Toderici et al., “Bidirectional relighting for 3D-aided 2D Face
`Recognition,” Proceedings from IEEE Computer Vision and Pattern
`Recognition, 13-18 June 2010, San Francisco, CA.
`Transcript of May 16, 2017, Deposition of Dr. Kakadiaris
`
`GTL
`Exhibit No.
`1014
`
`1015
`
`1016
`
`1017
`
`1018
`
`1019
`
`1020
`1021
`
`1022
`
`
`
`- ii -
`
`

`

`TABLE OF CONTENTS
`
`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`
`2.
`
`3.
`
`B.
`
`C.
`
`B.
`
`B.
`
`C.
`
`
`I.
`
`II.
`
`V.
`
`Securus’s anti-spoofing arguments fail. .......................................................... 3
`A.
`“Actual face” doesn’t mean “the user’s physical face and not a
`facsimile of a face such as a photograph.” ............................................ 3
`1.
`The ’420 patent’s specification and claims support Dr.
`Beigi’s, not Securus’s, interpretation. ......................................... 4
`Plain meaning supports Dr. Beigi’s, not Securus’s,
`interpretation. .............................................................................. 6
`The ’420 patent’s prosecution history supports Dr.
`Beigi’s, not Securus’s, interpretation. ......................................... 7
`Even under Securus’s construction of “actual,” the art will
`verify that the physical face is present in the image as claimed. .......... 9
`“Feature detection process” in the ’420 patent does not require
`anti-spoofing. .......................................................................................12
`The art requests to initiate an electronic visitation session in just the
`manner disclosed in the ’420 patent. .............................................................14
`A.
`Torgersrud’s request is indistinguishable from the one disclosed
`in the ’420 patent. ................................................................................16
`Kenoyer discloses no input at all between the request and
`capture of the image. ...........................................................................20
`III. Both Torgersrud and Kenoyer disclose feature detection, and Zhang
`suggests making the feature detection three-dimensional as claimed. ..........20
`A.
`Zhang teaches that its three-dimensional techniques can be used
`with two-dimensional images..............................................................21
`At the generality claimed, a skilled artisan is capable of
`incorporating three-dimensional techniques into the system of
`Torgersrud and Kenoyer. .....................................................................22
`A skilled artisan would recognize that incorporating three-
`dimensional techniques has a predictable result of being more
`accurate. ...............................................................................................24
`IV. The Petition fully addressed claim 21, showing it to be obvious over
`Torgersrud, Kenoyer, and Zhang. ..................................................................26
`Conclusion .....................................................................................................27
`
`
`
`- iii -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`The ’420 patent relates to video conferencing with inmates of a correctional
`
`facility, such as a prison. In the ’420 patent, Securus claims to have invented, just
`
`three years ago, face recognition applied during a video conference with an inmate.
`
`At institution, the Board found a reasonable likelihood that all claims of the ’420
`
`patent are obvious over Torgersrud. In particular, the Board instituted review of
`
`claims 1-9 and 11-19 over Torgersrud and Kenoyer, and claims 10, 20, and 21 over
`
`Torgersrud, Kenoyer, and Zhang. Facing these grounds, Securus makes four
`
`arguments.
`
`First, Securus claims for the first time that its real innovation isn’t face
`
`recognition, but “the ability to distinguish ‘actual faces’ from imitations, such as
`
`photographs” by evaluating changes between video frames. (Ex. 2004, ¶45.)
`
`Securus calls this ability anti-spoofing. In making this argument, Securus seeks to
`
`rewrite its claims, advancing two new constructions. Securus’s first construction is
`
`both incorrect and ineffectual. Securus argues that the claimed “actual face” is “the
`
`user’s physical face and not a facsimile of a face such as a photograph.” (Paper 17
`
`(“POR”), 11.) The ’420 patent specification doesn’t define “actual.” In fact, none
`
`of the evidence that Securus cites—the ’420 patent’s specification and prosecution
`
`history, and a lay dictionary—supports its interpretation that an “actual” face is in
`
`contrast to a “facsimile.” Instead, each of those sources support the understanding
`
`of Dr. Beigi—GTL’s expert with 20 years’ experience in facial detection and
`
`
`
`- 1 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`recognition (Ex. 1006, ¶¶11-14)—that, in the context of the ’420 patent, the word
`
`“actual” distinguishes between “[a] face and… something of its like.” (Ex. 2006,
`
`120:23.) However, even Securus’s construction of “actual” doesn’t overcome the
`
`art, because Torgersrud verifies the presence of a physical face, that is, the
`
`inmate’s face, before allowing login. Apparently realizing that Securus’s
`
`construction of “actual” doesn’t save the ’420 claims, Securus also construes
`
`“feature detection process” to require “calculating changes in the measurements in
`
`the sequence of video frames or still images.” (POR, 20.) That construction is
`
`absurdly narrow, defying basic claim construction principles. (Paper 8 (“ID”), 14-
`
`15.)
`
`Second, Securus argues a request cited from Torgersrud isn’t a “request to
`
`initiate an electronic visitation session” as claimed. (POR, 41-42.) Securus’s
`
`argument fails, because Torgersrud’s request is indistinguishable from the ’420
`
`patent’s. And, regardless, Kenoyer’s request, which the Petition also cited, satisfies
`
`what Securus apparently thinks is required of the claimed “request to initiate an
`
`electronic visitation session.”
`
`Third, Securus raises various concerns with Zhang as applied to the ’420
`
`patent’s claims 10, 20, and 21, which relate to three-dimensional feature detection.
`
`For example, Securus contends that a skilled artisan couldn’t make the
`
`
`
`- 2 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`combination claimed. Here, Securus ignores what Zhang teaches and vastly
`
`overstates what a skilled artisan would need to implement the claims.
`
`Fourth and finally, Securus takes issue with how the Petition treats claim 21,
`
`referring to discussion of nearly identical language from in claim 1. (POR, 52-54.)
`
`Securus’s argument fails, because the Petition addresses every claim feature,
`
`explaining how the prior art renders each obvious.
`
`I.
`
`Securus’s anti-spoofing arguments fail.
`“Actual face” doesn’t mean “the user’s physical face and not a
`A.
`facsimile of a face such as a photograph.”
`
`Securus alleges that the term “actual face” in the claims should be construed
`
`as “the user’s physical face and not a facsimile of a face such as a photograph.”
`
`(POR, 11.) For this construction, Securus points to the ’420 specification and
`
`claims, a dictionary definition, and the ’420 patent’s prosecution history. Securus’s
`
`construction fails, because none of these sources limits, or even describes, the word
`
`“actual” to be in contrast with a “facsimile.” In context of the relevant art and the
`
`’420 patent, “actual” is used to distinguish between those regions of an image that
`
`include a face and those that might. Dr. Beigi succinctly explained that the word
`
`“actual” in the phrase “actual face” is meant to distinguish between “[a] face and…
`
`something of its like.” (Ex. 2006, 120:23.) Dr. Beigi explained “there are things
`
`that may look like faces because they have features that are similar. For example,
`
`there may be two circles and, you know, something that looks like a mouth or
`
`
`
`- 3 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`something that looks like an oval which is not really an actual face.” (Ex. 2006,
`
`120:23-121:4.) Dr. Beigi’s understanding is supported by each piece of evidence
`
`that Securus cites.
`
`The ’420 patent’s specification and claims support Dr.
`1.
`Beigi’s, not Securus’s, interpretation.
`
`Starting with the specification, Securus doesn’t contend that it acted as its
`
`own lexicographer to define “actual.” Nor could it because “the ’420 patent,” the
`
`Board correctly recognized at institution, “does not provide a definition of the term
`
`‘actual face.’” (ID, 29.) Instead, Securus points to a single sentence: “[i]n certain
`
`embodiments, changes in the measurements from frame to frame of a video image
`
`may indicate that the person is [1] an actual person and [2] not simply a
`
`photograph presented to trick the system.” (Ex. 1001, 9:5-8.) The sentence doesn’t
`
`describe an actual person as being “not simply a photograph presented to trick the
`
`system.” Far from equating the two concepts, the sentence lists them as separate
`
`modifiers, describing how the “person” is both “actual” and “not... a photograph
`
`presented to trick the system.” Securus’s interpretation conflicts with the general
`
`presumption that every word in a patent adds meaning. See Merck & Co., Inc. v.
`
`Teva Pharm. USA, Inc., 395 F.3d 1364, 1372 (Fed. Cir. 2005). If Securus’s claim
`
`that the ’420 patent uses word “actual” to distinguish from photographs was true,
`
`then the phrase “and not simply a photograph presented to trick the system” would
`
`be redundant. See id. Outside of the one sentence that Securus cites, the
`
`
`
`- 4 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`surrounding specification disclosure uses the word “actual” at least nine times. (Ex.
`
`1001, Abstract, FIG. 4 (403, 404), 1:50, 1:57, 1:59, 9:1, 9:15, 10:15.) Yet in no
`
`instance is the word “actual” used to distinguish from “a facsimile... such as a
`
`photograph.”
`
`Lacking any specification definition of the claimed “actual face,” Securus
`
`turns to claim differentiation, arguing: “Independent claim 21… uses the claim
`
`term ‘face,’ thus creating a presumption that the term ‘face’ has a broader scope
`
`than the term ‘actual face.’” (POR, 12.) However, as noted above, GTL never
`
`contended that “face” and “actual face” are synonymous. Moreover, claim 21 has
`
`other differences with claims 1 and 11 that give it a distinct scope.
`
`The context provided by the surrounding claims, in fact, contradicts
`
`Securus’s construction. As described in more detail below in Section III, the
`
`construction of Securus’s declarant Dr. Kakadiaris requires that a “feature
`
`detection process” be applied to several images. (See Ex. 2004, ¶96.) Dr.
`
`Kakadiaris confirmed this understanding at deposition, when he was unable to
`
`identify any anti-spoofing technique that could be conducted with a single image.
`
`(Ex. 1022, 11:14-17.) Yet, that “performing” step of claim 1 only requires that the
`
`feature detection process be performed on a single image. (See also Ex. 1001,
`
`Claims 3 and 4 (reciting a singular “still-frame image” or “video frame”).) Needing
`
`
`
`- 5 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`only a single frame, the claim language contradicts Securus’s construction of
`
`“actual” as meaning anti-spoofing.
`
`Plain meaning supports Dr. Beigi’s, not Securus’s,
`2.
`interpretation.
`
`Lacking any definition of “actual” from the specification and the claims,
`
`Securus cherry picks a definition from a non-technical dictionary that doesn’t even
`
`support the position it advances. (POR, 16.) That definition—the third of at least
`
`six listed—doesn’t contrast with a “facsimile such as a photograph.” (Ex. 2005,
`
`22.) Instead, that definition only describes “actual” as being “not spurious, REAL,
`
`GENUINE.” (Id.) That definition—defining “actual” as “real”—is, in fact, more
`
`consistent with Dr. Beigi’s interpretation distinguish between “[a] face and…
`
`something of its like.” (Ex. 2006, 120:23-121:4.) As Dr. Beigi explained, an image
`
`can include “things that may look like faces because they have features that are
`
`similar,” but aren’t facsimiles, such as photographs. (Id.)
`
`The other definitions that Securus’s dictionary lists also fail to contrast with
`
`a facsimile and instead support Dr. Beigi’s interpretation. For example, before
`
`Securus’s chosen definition, the dictionary describes “actual” as meaning “existing
`
`in act... contrasted with potential and possible” and “existing in fact... distinguished
`
`from apparent and nominal.” (Ex. 2005, 22.) Hence, when applied to the context of
`
`face detection and recognition, the word “actual,” is meant to distinguish those
`
`regions of the image that are potentially or apparently a face, what Dr. Beigi
`
`
`
`- 6 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`described as “things that may look like faces because they have features that are
`
`similar.” (Ex. 2006, 120:23-121:4.)
`
`Dr. Beigi’s understanding of “actual” is consistent with how Dr. Kakadiaris
`
`has used the word in his own writings related to face detection and recognition. In
`
`one of his publications, Dr. Kakadiaris describes an “actual geometry of the face.”
`
`(Ex. 1021, 260.) He acknowledged that this paper relates to face recognition (Ex.
`
`2022, 27:6) and that neither this paper nor any of his other publications relate to
`
`anti-spoofing (id., 20:1-5). Hence, plain meaning supports Dr. Beigi’s, not
`
`Securus’s, interpretation of “actual.”
`
`The ’420 patent’s prosecution history supports Dr. Beigi’s,
`3.
`not Securus’s, interpretation.
`
`Securus next argues the ’420 patent’s prosecution history, even alleging
`
`prosecution disclaimer of claim scope. (POR, 17.) Time and again, the Federal
`
`Circuit has “declined to apply the doctrine of prosecution disclaimer where the
`
`alleged disavowal of claim scope is ambiguous.” See, e.g., Omega Engineering,
`
`Inc., v. Raytek Corp., 334 F.3d 1314, 1324 (Fed. Cir. 2003). Such a disclaimer
`
`applies only “where the patentee has unequivocally disavowed a certain meaning
`
`to obtain his patent.” Id. No such unambiguous, unequivocal disavowal exists here.
`
`In fact, contrary to Securus’s allegations, anti-spoofing was never mentioned
`
`during the ’420 patent’s prosecution. Securus never argued that an “actual face” is
`
`
`
`- 7 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`“the user’s physical face and not a facsimile of a face such as a photograph,” or
`
`even characterized the word “actual” at all in the phrase “actual face.”
`
`In the first office action, the examiner rejected the claims as being obvious
`
`over two references: Keiser (U.S. Pat. Pub. No. 2008/0000966) and Laney (U.S.
`
`Pat. Pub. No. 2012/0281058). (Ex. 1002, 0068.) In response to the first office
`
`action, Securus merely stated that “Keiser discloses capturing an image [of a
`
`visitor] and comparing it to another image, but fails to teach or suggest verifying
`
`that an actual face of an actual person is present in an image.” (Id., 0059.) All this
`
`states is that the content in the image Keiser compared isn’t necessarily a visitor’s
`
`face. Keiser doesn’t mention a face, and it could look to some other distinctive
`
`feature of the visitor in making a comparison. This argument doesn’t relate to anti-
`
`spoofing.
`
`Indeed, in the next, final action, the examiner provided no indication that he
`
`understood Securus’s arguments as referring to anti-spoofing. Instead, the
`
`examiner argued that, even though Keiser doesn’t explicitly mention a visitor’s
`
`face, by capturing an image of the visitor, Keiser’s visitation system also captures
`
`an image of the visitor’s face. (Ex. 1002, 48.) Further the examiner argued that, by
`
`recognizing the visitor, Keiser’s visitation system verifies presence of an actual
`
`face, the actual face of the visitor: “the visitation system captured the visitor
`
`image (i.e., actual face of the visitor) to verify that the visitor is the actual person
`
`
`
`- 8 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`that registered by comparing with stored data.” (Id. (emphasis in original).) Thus,
`
`the Examiner finally rejected the claims.
`
`After the final rejection, Securus, it now admits, tried to overcome the
`
`examiner’s interpretation by adding a negative limitation to exclude facial
`
`recognition, explicitly stating the feature detection process is performed “without
`
`performing a facial recognition process.” (POR, 19; Ex. 1002, 0035, 0039.) That
`
`amendment was never entered.1 (See Ex. 1002, 0021.) As discussed below in
`
`section I.C, by proposing an amendment instead of disputing the examiner’s
`
`interpretation, Securus acquiesced to it, effectively conceding that facial
`
`recognition, without anti-spoofing, teaches the “actual face” language claimed.
`
`Lacking any support from the specification, claims, dictionary evidence, and
`
`prosecution history, Securus’s construction of “actual face” as meaning “the user’s
`
`physical face and not a facsimile of a face such as a photograph” fails.
`
`Even under Securus’s construction of “actual,” the art will verify
`B.
`that the physical face is present in the image as claimed.
`
`Even given Securus’s construction of “actual face,” Torgersrud still
`
`
`1 Apparently, the examiner found the amendment unpersuasive. Instead, as
`
`explained in the Petition, the amendment that persuaded the Examiner recites a fea-
`
`ture that Securus doesn’t dispute is in Torgersrud. (Paper 2 (“Pet.”), 2-9.)
`
`
`
`- 9 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`suggests the claims, because Torgersrud, through its facial recognition
`
`functionality, “perform[s] a feature detection process... to verify that an actual face
`
`was present an image” as recited in independent claims 1 and 11.
`
`During Torgersrud’s normal operation, a resident presents its “physical face
`
`and not a facsimile of a face such as a photograph.” That so-called “actual” face is
`
`detected and “facial verification is performed.” (Ex. 1003, ¶66 (emphasis added).)
`
`To perform facial verification, “the image captured [of the resident’s physical face]
`
`is compared with “a pre-stored image of the resident,” the pre-stored image being
`
`the resident’s physical face. (Id., ¶67.) In particular, “[t]he facial recognition
`
`system uses facial ‘landmarks’ generated by mathematical formulas to present a
`
`score which indicates a likelihood that the captured image matches the pre-stored
`
`image. If the images match to a sufficient degree, the verification is approved.”
`
`(Id.) In this way, Torgersrud “perform[s] a feature detection process... to verify that
`
`an actual face was present an in image” as claimed.
`
`Disputing this reading, Securus argues that Torgersrud does not
`
`“distinguish[] whether a user is an ‘actual person and not simply a photograph
`
`presented to trick the system.’” (POR, 30.) But “distinguishing” is not recited in
`
`the claims. The ’420 specification describes “determin[ing] if the presented person
`
`is an actual person or a photograph.” (Ex. 1001, 9:8-12.) But that is not claimed.
`
`The claims recite “verifying that....,” not “determining if...” In particular, the
`
`
`
`- 10 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`claims recite “verify[ing] that an actual face was present an image claim.” Because
`
`that is how Torgersrud’s “facial verification” operates in its normal operation on
`
`the inmate’s face—a physical face—Torgersrud teaches the claims.
`
`This reasonable reading is how both the examiner and Securus interpreted
`
`the ’420 claims during prosecution. As noted above, during prosecution, the
`
`examiner found that facial recognition—verifying that the image depicts a face
`
`matching what was previously registered for a user—is within the scope of the
`
`claims. The examiner explained that the prior art’s “visitation system captur[es]
`
`the visitor image (i.e., actual face of the visitor) to verify that the visitor is the
`
`actual person that registered by comparing with stored data” (Ex. 1002, 0048
`
`(emphasis in original).) In response to that finding, for then claim 12 (now claim
`
`11), the only distinction Securus argued was a proposed amendment (which was
`
`not entered) to explicitly exclude facial recognition. (Ex. 1002, 0035, 0039.)
`
`Securus did not, as it does now, argue that “actual” connotes anti-spoofing. (Id.)
`
`By offering an amendment to explicitly exclude facial recognition, Securus
`
`effectively conceded that the claims as drafted before the amendment, and as now
`
`issued, cover facial recognition, which Torgersrud teaches. (Id., 0038-40; see also
`
`Ex. 1001, claim 5 (reciting facial recognition).)
`
`Thus, in performing facial recognition, Torgersrud verifies that an “actual”
`
`face is in the image, even under Securus’s interpretation of “actual.”
`
`
`
`- 11 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`“Feature detection process” in the ’420 patent does not require
`C.
`anti-spoofing.
`
`Realizing that construing the word “actual” alone will not save its claims
`
`over Torgersrud, Securus argues a still narrower construction. In particular, Dr.
`
`Kakadiaris argued that a “feature detection process” as recited in independent
`
`claims 1 and 11 is “a process for identifying characteristics of image, measuring
`
`facial landmarks at a plurality points on the image, calculating changes in the
`
`measurements in a sequence of video frames or still images, and determining
`
`whether the changes are indicative of actual face.” (Ex. 2004, ¶96; POR, 37-38.) In
`
`contrast, neither Dr. Beigi nor the Board read the claimed “feature detection
`
`process” to require that changes be calculated in a sequence of images, let alone to
`
`determine what the supposed changes calculate. The Board construed “feature
`
`detection process” to be “a process for detecting characteristics of an image.” (ID,
`
`12.) Dr. Beigi had a similar reading. (Ex. 1006, ¶66.)
`
`The Board supports its construction with description from the ’420
`
`specification: “A... feature detection process may identify... characteristics of an
`
`image, including measurements of features of a face at a plurality of points on the
`
`image.” (Ex. 1001, 9:2-5.) The ’420 patent’s usage, Dr. Beigi explained, is
`
`consistent with how the term is used in the art. (Ex. 1006, ¶27; see also ¶¶28-29,
`
`34, 36; see generally also Ex. 1007-1011.) Even Dr. Kakadiaris acknowledged that
`
`his construction departs from how the term “feature” is used elsewhere in the art.
`
`
`
`- 12 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`At deposition, Dr. Kakadiaris was asked what the difference was between his
`
`usage and the usage of a survey of face detection techniques that Dr. Beigi
`
`provided. (Ex. 1022, 65:13-16.) In response, Dr. Kakadiaris stated: “in this
`
`document, they don’t describe whether the changes or the features are indicative of
`
`an actual face.” (Id., 65:17-20.)
`
`To justify departure from how the term is understood in the art, Securus
`
`again only points to that single column 9 passage discussed above. (POR, 22.) That
`
`passage starts with the sentence above supporting the Board’s construction,
`
`describing that “A... feature detection process may identify... characteristics of an
`
`image.” (Ex. 1001, 9:2-5.) Following are the two sentences that Securus relies on
`
`for its construction. But, as this Board correctly recognized applying Federal
`
`Circuit precedent (ID, 15), “a particular embodiment appearing in the written
`
`description may not be read into a claim when the claim language is broader than
`
`the embodiment.” Superguide Corp. v. DirecTV Enterprises, Inc., 358 F.3d 870,
`
`875 (Fed. Cir. 2004). Here, both sentences are prefaced to explicitly state that the
`
`features they described are not present in all embodiments of the ’420 invention.
`
`The first sentence is prefaced with the phrase “[i]n certain embodiments” and the
`
`second sentence is similarly prefaced with “[i]n still another embodiment.” (Id.,
`
`9:5-11.) Thus, these embodiments should not be read into the claims.
`
`
`
`- 13 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`In sum, the “feature detection process” recited in the ’420 patent’s claims 1
`
`and 11 doesn’t require anti-spoofing, and the Board should maintain the
`
`construction of “feature detection process” made at institution. That construction,
`
`the Petition explains, is taught in Torgersrud. (Pet., 30-33.)
`
`II. The art requests to initiate an electronic visitation session in just the
`manner disclosed in the ’420 patent.
`
`Securus argues that the claimed “electronic visitation session” means “the
`
`period of time during which a resident of a controlled environment facility
`
`participates in an electronic communication with a nonresident visitor, wherein the
`
`session begins when the electronic communication is connected and ends when it
`
`is disconnected.” (POR, 29.) On institution, the Board adopted a somewhat broader
`
`construction. (ID, 12.) GTL disputes those constructions, and maintains the
`
`construction and arguments set forth in the Petition. (Pet., 22-23.) But even under
`
`Securus’s narrower construction, there’s no dispute that an “electronic visitation
`
`session” is taught in Torgersrud. This is because Torgersrud discloses voice calls
`
`(Ex. 1003, ¶76) and video conferences (Id., ¶22). Voice calls and video conference,
`
`even Securus acknowledges, have a beginning and an end as required by Securus’s
`
`construction of “electronic visitation session.”2 (POR, 29; Ex. 2004, ¶83.)
`
`
`2 Torgersrud also explicitly discloses the beginning and end in several con-
`
`texts, including providing “schedule limits defined per resident” (Ex. 1003, ¶¶45,
`
`
`
`- 14 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`Instead of arguing that the art lacks an electronic visitation session, Securus
`
`argues that Torgersrud lacks “capturing an image, with an image capture device, of
`
`a user in response to the request” where the request is the claimed “request to
`
`initiate an electronic visitation session.” (POR, 41-42.) Torgersrud, Securus
`
`acknowledges, captures an image in response to a resident’s selection of a login
`
`button. (Id.) Then, after login, Torgersrud provides a menu which includes options
`
`governed by permissions and is accessible using not only face recognition also
`
`using a PIN. (Id.) The resident selects one of the menu options to initiate the
`
`electronic visitation session. (Id.) Because Torgersrud provides a menu before
`
`initiating an electronic visitation session, the selection of the login button, Securus
`
`argues, is not a “request to initiate an electronic visitation session” as claimed. (Id.)
`
`Securus’s argument fails for two reasons. First, Torgersrud’s request is
`
`indistinguishable from the one disclosed in the ’420 patent. Second, the Petition
`
`also cited to Kenoyer for the claimed request, which Securus doesn’t even address
`
`in its POR.
`
`
`61, 63, 83, 84), “disconnect[ing] a conference when a second face is present” (id.,
`
`¶86), and limiting times based on these available in a prepaid account (id., ¶105).
`
`
`
`- 15 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`A. Torgersrud’s request is indistinguishable from the one disclosed
`in the ’420 patent.
`
`A claim interpretation that excludes the preferred embodiment is “rarely, if
`
`ever, correct.” Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1583 (Fed. Cir.
`
`1996). Here, an interpretation of the claimed “request to initiate electronic
`
`visitation session” that excludes Torgersrud’s login button would necessarily
`
`exclude the request disclosed in the ’420 patent.
`
`The relevant teachings of Torgersrud are essentially undisputed. “[A]
`
`resident can select the login icon to log in to the system” using a PIN. (Ex. 1003,
`
`¶60 (reference number omitted).) “During the log in process, the camera of the
`
`kiosk is activated. An image or video of the resident logging on is recorded.” (Id.,
`
`¶64 (reference number omitted).) The “kiosk will use facial detection software to
`
`ensure that a face is present in the camera field of view.” (Id.) In addition to a PIN,
`
`the kiosk uses facial recognition. (Id.) An interface prompting the facial
`
`recognition is illustrated in Torgersrud’s FIG. 9, reproduced below. (Id., ¶66.)
`
`“When the resident selects [a] verify icon 904 [on the interface], facial verification
`
`is performed.” (Id.) If an image captured by the camera of the kiosk matches a pre-
`
`stored image of the resident, “the resident is granted access to the system.” (Id.,
`
`¶67 (reference numbers omitted).)
`
`
`
`- 16 -
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`
`
`
`Once granted access, a resident is “presented with a main screen 1011…,
`
`illustrated in FIG. 10,” reproduced below. (Id., ¶68.) “The options available at the
`
`main screen 1011 can be configured by facility administrators.” (Id.) The options
`
`“can include phone calling,… text messaging, video conferencing, [and]
`
`commissary ordering.” (Id.)
`
`
`
`- 17 -
`
`
`
`

`

`Case IPR2016-01220 of
`U.S. Pat. No. 9,007,420
`When a user in Torgersrud desires an electronic visitation session, the first
`
`thing she’ll do is press the login button. To initiate an electronic visitation session,
`
`the user must f