`
`______________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`______________________________
`
`CANON INC. and CANON U.S.A., INC., and
`AXIS COMMUNICATIONS AB,
`
`Petitioners,
`v.
`
`AVIGILON FORTRESS CORPORATION,
`Patent Owner.
`______________________________
`
`Case: IPR2019-00311
`U.S. Patent No. 7,932,923
`______________________________
`
`PETITIONERS’ REPLY TO PATENT OWNER’S RESPONSE
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`TABLE OF CONTENTS
`
`2.
`
`3.
`
`B.
`C.
`D.
`
`B.
`
`C.
`
`INTRODUCTION ........................................................................................ 1
`I.
`COLLATERAL ESTOPPEL APPLIES .................................................... 2
`II.
`III. CLAIM CONSTRUCTIONS....................................................................... 4
`A. MEANS-PLUS-FUNCTION ELEMENTS (CLAIMS 9-19, 30-
`41) ......................................................................................................... 4
`“ATTRIBUTES” LIMITATIONS (CLAIMS 1-41) ............................ 5
`“NEW USER RULE” (CLAIMS 1-41) ............................................... 5
`INDEPENDENCE-BASED LIMITATIONS (CLAIMS 1-41) ........... 6
`1.
`INDEPENDENCE ARGUMENT (1) (AVIGILON’S
`“APPLYING” ARGUMENT) ................................................... 6
`INDEPENDENCE ARGUMENT (3) (AVIGILON’S
`“EVENT” ARGUMENT) .......................................................... 7
`INDEPENDENCE ARGUMENT (2) (AVIGILON’S
`“INDEPENDENT” ARGUMENT) ........................................... 8
`THE “ONLY” LIMITATIONS (CLAIMS 1-41) ................................ 9
`E.
`“VIDEO DEVICE” ............................................................................ 10
`F.
`IV. THE CHALLENGED CLAIMS ARE UNPATENTABLE .................... 10
`A.
`KELLOGG DISCLOSES THE “SINGLE CAMERA”
`LIMITATIONS .................................................................................. 10
`KELLOGG DISCLOSES “SELECTING A NEW USER RULE
`AFTER DETECTING THE PLURALITY OF ATTRIBUTES” ...... 13
`KELLOGG DISCLOSES “APPLYING THE NEW USER
`RULE TO THE PLURALITY OF DETECTED
`ATTRIBUTES” .................................................................................. 14
`KELLOGG DISCLOSES “APPLYING THE NEW USER
`RULE TO ONLY THE PLURALITY OF DETECTED
`ATTRIBUTES” .................................................................................. 16
`KELLOGG DISCLOSES “THE PLURALITY OF
`ATTRIBUTES THAT ARE DETECTED ARE
`INDEPENDENT OF WHICH EVENT IS IDENTIFIED” ............... 18
`
`D.
`
`E.
`
`i
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`F.
`
`I.
`
`J.
`
`G.
`H.
`
`KELLOGG DISCLOSES “SELECTING THE NEW USER
`RULE COMPRISES SELECTING A SUBSET OF THE
`PLURALITY OF ATTRIBUTES FOR ANALYSIS” ....................... 19
`KELLOGG DISCLOSES “A VIDEO DEVICE” ............................... 19
`KELLOGG DISCLOSES MEMORY “CONFIGURED TO
`STORE AT LEAST SOME OF THE PLURALITY OF
`ATTRIBUTES FOR AT LEAST TWO MONTHS” ......................... 20
`BRILL DISCLOSES THE “SINGLE CAMERA”
`LIMITATIONS .................................................................................. 21
`BRILL DISCLOSES “APPLYING THE NEW USER RULE
`TO ONLY THE PLURALITY OF DETECTED
`ATTRIBUTES” .................................................................................. 21
`A POSITA WOULD HAVE BEEN MOTIVATED TO
`COMBINE KELLOGG AND BRILL ................................................. 22
`AVIGILON’S “OBJECTIVE INDICIA OF NON-
`OBVIOUSNESS” IS NOT CREDIBLE ............................................ 23
`KELLOGG AND BRILL ARE PRINTED PUBLICATIONS ................. 24
`A.
`THE BOARD MAY CONSIDER POST-PETITION
`EVIDENCE CONFIRMING PUBLICATION .................................. 25
`THE EVIDENCE SUBMITTED WITH THE PETITION
`ALONE PROVES PUBLICATION .................................................. 26
`VI. CONCLUSION ........................................................................................... 28
`
`K.
`
`L.
`
`B.
`
`V.
`
`ii
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`LIST OF EXHIBITS
`
`Ex. 1001 U.S. Patent No. 7,932,923 (“the ’923 Patent”)
`Ex. 1002
`Prosecution History of the ’923 Patent
`Ex. 1003
`“Visual Memory” by Christopher James Kellogg (“Kellogg”)
`Ex. 1004
`“Event Recognition and Reliability Improvements for the
`Autonomous Video Surveillance System” by Frank Brill et al.
`(“Brill”)
`Ex. 1005 Declaration of John R. Grindon, D.Sc.
`Ex. 1006
`“Motion Recovery for Video Content Classification” by N. Dimitrova
`et al. (“Dimitrova”)
`Ex. 1007 Declaration of Emily R. Florio
`Ex. 1008
`February 29, 2012 Request for inter partes Reexamination of the ’923
`Patent
`Ex. 1009 May 23, 2012 Order Granting/Denying Request for inter partes
`Reexamination of the ’923 Patent
`Ex. 1010 May 23, 2012 Office Action in inter partes Reexamination of the ’923
`Patent
`Ex. 1011 August 27, 2012 Amendment and Reply in inter partes
`Reexamination of the ’923 Patent
`February 13, 2013 Decision Granting Petition to Terminate inter
`partes Reexamination Proceeding of the ’923 Patent
`Ex. 1013 May 23, 2013 Attachment to Request for ex parte Reexamination of
`the ’923 Patent
`June 17, 2013 Order Granting/Denying Request for ex parte
`Reexamination of the ’923 Patent
`Ex. 1015 August 30, 2013 Office Action in ex parte Reexamination of the ’923
`Patent
`
`Ex. 1012
`
`Ex. 1014
`
`iii
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Ex. 1016 October 30, 2013 Amendment and Reply in ex parte Reexamination
`of the ’923 Patent
`Ex. 1017 April 4, 2014 Final Office Action in ex parte Reexamination of the
`’923 Patent
`Ex. 1018 April 16, 2014 Amendment and Reply in ex parte Reexamination of
`the ’923 Patent
`Ex. 1019 April 30, 2014 Notice of Intent to Issue ex parte Reexamination
`Certificate of the ’923 Patent
`Ex. 1020 May 21, 2014 ex parte Reexamination Certificate of the ’923 Patent
`Ex. 1021 U.S. Patent No. 5,969,755 to Courtney (“Courtney”)
`Ex. 1022
`“Object-Oriented Conceptual Modeling of Video Data” by Young
`Francis Day et al., (“Day-I”)
`Ex. 1023 Declaration of Christopher James Bailey-Kellogg in IPR2018-00138
`and IPR2018-00140
`Ex. 1024 Applicant Response of June 11, 2012 in inter partes Reexamination of
`U.S. Patent No. 7,868,912 (“the ’912 Patent”)
`Ex. 1025 Decision Granting Petition to Terminate inter partes Reexamination
`of the ’912 Patent
`Ex. 1026 May 24, 2013 Attachment to Request for ex parte Reexamination of
`the ’912 Patent
`June 20, 2013 Order Granting/Denying Request for ex parte
`Reexamination of the ’912 Patent
`Ex. 1028 August 30, 2013 Office Action in ex parte Reexamination of the ’912
`Patent
`Ex. 1029 October 30, 2013 Amendment and Reply in ex parte Reexamination
`of the ’912 Patent
`Ex. 1030 March 27, 2014 Final Office Action in ex parte Reexamination of the
`’912 Patent
`
`Ex. 1027
`
`iv
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Ex. 1031 April 16, 2014 Amend and Reply in ex parte Reexamination of the
`’912 Patent
`Ex. 1032 Notice of Intent to Issue ex parte Reexamination Certificate of the
`’912 Patent
`Ex. 1033 Declaration of Kenneth A. Zeger for the ex parte Reexamination of
`the ’923 Patent
`Ex. 1034 U.S. Patent No. 7,868,912
`Ex. 1035 U.S. Patent No. 8,564,661
`Ex. 1036 Curriculum Vitae of John R. Grindon, D.Sc.
`Ex. 1037 Merriam-Webster’s Collegiate Dictionary
`Ex. 1038 Declaration of Dr. Alan Bovik in IPR2018-00138 and IPR2018-00140
`Ex. 1039
`The New York Times October 2, 1999, Pro Basketball; McHale and
`Thompson Enter Hall of Fame with 3 Others, by Mike Wise
`Ex. 1041 Declaration of Guang-Yu Zhu
`Ex. 1042
`“Understanding MARC Bibliographic: Parts 1 to 6,” available at
`https://www.loc.gov/marc/umb/um01to06.html (last accessed August
`1, 2019)
`“Understanding MARC Bibliographic: Parts 7 to 10,” available at
`https://www.loc.gov/marc/umb/um07to10.html (last accessed August
`1, 2019)
`Ex. 1044 Nevenka Dimitrova et al., “Motion Recovery for Video Content
`Classification,” ACM Transactions on Information Systems, Vol. 13,
`No. 4, 408-439 (1995) from the MIT Libraries (Served But Not Filed)
`Ex. 1045 Bruce E. Flinchbaugh et al., “Autonomous Video Surveillance,” in
`Emerging Applications of Computer Vision, Vol. 2962, pp. 144-151
`(1997) from the Library of Congress
`
`Ex. 1043
`
`v
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Ex. 1046
`
`Frank Brill et al., “Event Recognition and Reliability Improvements
`for the Autonomous Video Surveillance System” Proceedings of the
`Image Understanding Workshop, Nov. 20-23, 1998, Vol. 1, pp. 267-
`283 from the University of Virginia Library
`Ex. 1047 Declaration of Katie Zimmerman filed in KAZ USA, Inc. v. Exergen
`Corp., Case IPR2016-01437, Exhibit 1027
`Ex. 1048 Declaration of Marilyn McSweeney filed in Yahoo! Inc. v. CreateAds
`LLC, Case IPR2014-00200, Exhibit 1007
`Ex. 1049 Declaration of Bryan Patrick Kasik
`Ex. 1050
`2019.08.09 Conference Call Transcript
`Ex. 1051
`Frank Brill et al., “Event Recognition and Reliability Improvements
`for the Autonomous Video Surveillance System” Proceedings of the
`Image Understanding Workshop, Nov. 20-23, 1998, Vol. 1, pp. 267-
`283 from the University of Wisconsin-Madison Memorial Library
`Email from the University of Wisconsin-Madison Memorial Library
`Ex. 1052
`Ex. 1053 Declaration of Rachel Watters
`Ex. 1054 Declaration of Katherine Zimmerman relating to Kellogg (Pending
`Authorization)
`Ex. 1055 Declaration of Katherine Zimmerman relating to Dimitrova (Served
`But Not Filed)
`Transcript of Deposition of Dr. Alan Bovik
`
`Ex. 1056
`
`vi
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`I.
`
`INTRODUCTION
`
`Avigilon’s Response fails to rebut Petitioner’s showing of invalidity of the
`
`’923 Patent claims for two primary reasons.
`
`First, most of Avigilon’s arguments must be rejected as collaterally estopped
`
`because they have already been considered and rejected in the Final Written
`
`Decisions involving Avigilon’s related U.S. Patent No. 8,564,661.1 Papst
`
`Licensing GmbH & Co. KG v. Samsung Electronics America, 924 F.3d 1243,
`
`1250-51 (Fed. Cir. 2019).
`
`For example, Avigilon again argues that Kellogg is deficient because it does
`
`not show the “independence based claim elements” and rather merely discloses an
`
`event querying/filtering system, like the Courtney reference applied against the
`
`patent in reexamination.
`
`But Avigilon made the same argument in the ’661 IPR and the Board
`
`unequivocally rejected it there. See IPR2018-00138 FWD, 11.
`
`Similarly, Avigilon again argues that the claimed new user rule requires a
`
`response, which the Board also rejected. Id., 13.
`
`Second, Avigilon’s arguments here fail on their merits as discussed below.
`
`1 As the Board has already recognized (Paper 13, 3) the ’923 patent here is related
`to the ’661 patent.
`
`1
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`II.
`
`COLLATERAL ESTOPPEL APPLIES
`
`“Collateral estoppel protects a party from having to litigate issues that have
`
`been fully and fairly tried in a previous action and adversely resolved against a
`
`party-opponent.”2 Nestle USA, Inc. v. Steuben Foods, Inc., 884 F.3d 1350, 1351
`
`(Fed. Cir. 2018); Papst, 924 F.3d at 1250. Collateral estoppel applies to issues
`
`decided in IPRs, including issues decided during the invalidation of a related
`
`patent. Papst, 924 F.3d at 1249; Mobile Tech, Inc. v. Invue Security Product,
`
`IPR2018-00481, FWD, Paper 29 (July 16, 2019).
`
`After the filing of the current petition, the Board issued its Final Written
`
`Decisions invalidating the related ’661 patent. Avigilon withdrew its appeals of
`
`those decisions and is now collaterally estopped as to the issues the Board has
`
`already decided.
`
`Collateral estoppel bars Avigilon from re-litigating factual and legal issues
`
`here where: (1) the issue is the same as the issue in the prior action; (2) the issue
`
`was actually litigated in the first action; (3) there was a final judgement in the first
`
`action that necessarily required determination of the identical issue; and (4) the
`
`prior action featured full representation of the estopped party. Id. at 10. All four
`
`elements apply here.
`
`2 All emphasis throughout added unless otherwise stated.
`
`2
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`One, most of the issues raised by Avigilon in these proceedings are the same
`
`as issues the Board rejected in the ’661 patent Final Written Decision. The ’661
`
`and ’923 patents, claim priority to the same parent application and use the same
`
`claim terms. The relevant portions of the specifications and figures are identical.
`
`And Avigilon agrees the claim terms should be construed consistently between the
`
`patents. See, Paper 9 at 8 (relying on the ’661 IPR institution decisions); 9 (relying
`
`on ’661 IPRs for “rules”); 12 (same for event and independent terms); 13, 14, 35.
`
`As explained below (see Section III), Avigilon reargues the same claim
`
`construction positions that the Board has already rejected. And Avigilon does not
`
`argue3 that there is anything different about the ’923 patent to support a different
`
`result here. See, Nestle, 884 F.3d at 1351 (estoppel applies against similar claim
`
`terms from related patents). Similarly, Avigilon also improperly argues the Board
`
`should reject the same prior art, Kellogg and Brill, that it previously considered in
`
`the ’661 IPRs. The Board need not and should not reevaluate its prior invalidity
`
`analysis regarding the significance of this prior art.
`
`Two, not only were the issues the same as in the prior ’661 patent
`
`proceedings, but they were actually litigated in those proceedings. Specifically,
`
`claim construction and the content of the prior art were, obviously, the central
`
`3 Nor could Avigilon argue that the constructions should differ, having advocated
`for consistent constructions in its preliminary response.
`
`3
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`issues in the ’661 IPR and Avigilon fully argued for the patentability of the ’661
`
`patent. See Papst, 924 F.3d at 1252-53.
`
`Three, the Board’s final written decisions in the ’661 IPR proceedings
`
`became irrevocably final when Avigilon dropped its appeals. Papst, 924 F.3d at
`
`1249 (citations omitted). These decisions were also necessary to the ’661 IPR final
`
`judgment. Claim construction is a prerequisite for determining validity. Oakley,
`
`Inc. v. Sunglass Hut Int’l, 316 F.3d 1331, 1339 (Fed. Cir. 2003). The Board, of
`
`course, necessarily rejected each of Avigilon’s validity arguments regarding the
`
`scope and content of the prior art when it found the ’661 patent claims invalid.
`
`Four, Avigilon was fully represented and vigorously litigated the prior
`
`action. In that action, Avigilon presented testimony from Dr. Bovik, cross
`
`examined Petitioner’s expert Dr. Grindon, and argued at the oral hearing. Both
`
`experts are the same as in this proceeding. On that full record, the Board issued a
`
`thorough, well-reasoned opinion.
`
`Elements one through three are specifically addressed below for each
`
`collaterally estopped issue. Element four is the same for all issues.
`
`III. CLAIM CONSTRUCTIONS
`
`A. Means-Plus-Function Elements (Claims 9-19, 30-41)
`
`The Board should adopt the Petitioner’s proposed constructions, which
`
`Avigilon “does not challenge.” Paper 27, 15.
`
`4
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`B.
`
`“Attributes” Limitations (Claims 1-41)
`
`Avigilon summarily states that “[n]o construction of ‘attributes’ is necessary
`
`here for Petitioners’ challenges,” conceding that the construction of these claim
`
`terms is not material to its validity arguments. Id., 6.
`
`C.
`
`“New User Rule” (Claims 1-41)
`
`The Board construed the substantively identical “rule” limitations in the
`
`related ’661 IPRs to not require a response. See IPR2018-00138 FWD, 11-13
`
`(“[a]lthough Patent Owner argues that a rule requires more than a query that
`
`returns whether an event has occurred (PO Resp. 31–32; Sur-Reply 9–10), we
`
`agree with Petitioner that a ‘response’ is not required”); IPR2018-00140 FWD, 6
`
`(same construction).
`
`Avigilon raises the same issue here, arguing that a “new user rule” requires a
`
`response. Paper 27, 6-8. Nothing about the ’661 patent makes this issue different.
`
`Because this same issue was actually litigated and necessary to the Board’s final
`
`judgment in the ’661 IPRs, collateral estoppel applies.
`
`Moreover, the ’923 patent explains that rules and responses are separate
`
`elements, and that responses are optional. See, e.g., Ex. 1001, 8:37 (“In block 34,
`
`a response is optionally identified”); 8:50-58 (“responses identified in block 34
`
`are optionally associated with each event discriminator”); 11:13-16.
`
`5
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`D.
`
`Independence-Based Limitations (Claims 1-41)
`
`1.
`
`Independence Argument (1) (Avigilon’s “Applying”
`Argument)
`
`Avigilon presents no evidence that supports its assertion that the “plain and
`
`ordinary” meaning of “applying the new user rule to the plurality of detected
`
`attributes” requires some special, undisclosed “analysis” on top of a “query.”
`
`Paper 27, 8-10. And, it offers no dictionary definitions or other evidence showing
`
`that the plain meanings of these words necessitate these additional limitations.
`
`Avigilon’s attempt to incorporate some particular “analysis” into the word
`
`“applying” is belied by the claims themselves, which use the terms differently. For
`
`example, claim 1 uses the term “applying the new user rule,” in the present
`
`limitation, but also recites “analyzing the video.”
`
` Moreover, “querying” is not merely “retrieving the detected attributes from
`
`a database” as Avigilon suggests. Id., 9. Rather, a “query” must examine the
`
`attributes to find a match for the collection of attributes specified by the user rule.
`
`Avigilon argues that the use of “analysis/analyzing” in dependent claims 2
`
`and 13 requires that the independent claims include those limitations is illogical.
`
`Dependent claims are narrower than the claims from which they depend. Thus
`
`reciting “analysis” in a dependent claim does not work backwards to require
`
`analysis in the independent claim.
`
`6
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`The ’923 patent gives no special meaning to the term “applying.” Indeed,
`
`the specification never once uses “applying” as the claim does. Nor does it suggest
`
`that “applying” should be limited to requiring analysis. The generic statement that
`
`the system can “further analyze previously processed video without needing to
`
`reprocess completely the video,” relied on by Avigilon (Paper 27, 10) is
`
`completely consistent with “querying” stored attributes, which does not require
`
`reprocessing the video.
`
`Avigilon’s expert witness, Dr. Bovik, admitted that “querying” meets the
`
`“applying” limitation: “A POSITA would understand that the claim limitation
`
`‘applying the new user rule to only the plurality of detected attributes’ means a
`
`system that can search, query, or analyze.” Ex. 2019, 42. Dr. Bovik’s
`
`unsupported statements about passive vs. active analysis (Paper 27, 10) do not
`
`contradict this admission.
`
`Finally, Avigilon’s reference to a statement that “a decision is made” (id, 10)
`
`in the ’923 patent reexamination does not reference analysis or constitute a clear an
`
`unmistakable disclaimer.
`
`2.
`
`Independence Argument (3) (Avigilon’s “event” Argument)
`
`Avigilon provides no response and only improperly incorporates “the
`
`reasons stated in” its Preliminary Response (id., 10) by reference, which violates
`
`7
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`37 CFR § 42.6(a)(3) and the Scheduling Order. Paper 14, 7. Petitioner’s
`
`construction should be adopted.
`
`3.
`
`Independence Argument (2) (Avigilon’s “independent”
`Argument)
`
`Avigilon first states that no construction is necessary, but then repeats its
`
`same failed argument from the ’661 IPRs that “independent” means “the attributes
`
`are detected without regard to or knowledge of events or identification of events.”
`
`Paper 27, 11-12. The Board has already rejected this same argument and ruled that
`
`there is no reason to so limit the claims, in ruling on this necessarily decided issue.
`
`See IPR2018-00138 FWD, 9. As such, collateral estoppel applies.
`
`Notwithstanding Avigilon’s attempt to characterize this as a new
`
`construction due to its addition of the phrase “or identification of events” (Paper
`
`27, 11-12), this is the same already, finally rejected argument. Indeed, Avigilon
`
`calls it a “minor alteration.” Id. Moreover, the addition of a second set of
`
`information (the identification of events) is baseless. Avigilon provides no citation
`
`to the ’923 patent specification nor any other explanation regarding what it means
`
`to detect attributes “without regard to or knowledge of …identification of events.”
`
`Its construction should be rejected and Petitioners’ unrebutted construction should
`
`be adopted.
`
`8
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`E.
`
`The “Only” Limitations (Claims 1-41)
`
`Avigilon’s argument that the claims include a negative limitation precluding
`
`the searching of abstractions (Paper 27, 14) must fail because the ‘923 patent
`
`provides no written description support for that limitation. Inphi Corp. v. Netlist,
`
`Inc., 805 F.3d 1350, 1355 (Fed. Cir. 2015). The ’923 patent does not disclose a
`
`system that never searches abstractions. Even if it did, the ’923 Patent does not
`
`disclose any benefits or reasons to avoid searching abstractions. To the contrary,
`
`all the embodiments of the ’923 patent state that both primitives (attributes) and
`
`abstractions may be searched. Ex. 1001, 8:16-17, 8:50-52; see Ex. 1056, 156:17-
`
`158:9 (Avigilon’s expert admitting abstractions “might be involved in” “defining
`
`events”). A construction of the claims forbidding searching abstractions lacks
`
`written description support in the ’923 patent and would be an error.
`
`Avigilon’s reliance on Crystal Semiconductor Corp. v. TriTech
`
`Microelectronics Int’l, Inc., 246 F.3d 1336, 1348 (Fed. Cir. 2001) is misplaced
`
`because that case explains that “comprising” creates a presumption that “the claim
`
`does not exclude additional, unrecited elements.” There is no claim or
`
`specification support for the argument that a system must completely exclude
`
`searches on abstractions to comport with the claims. Thus, at most, this limitation
`
`only requires an embodiment that can search attributes alone. See Ex. 1056,
`
`9
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`207:16-208:15 (if a system “practices every element of the claim,” it is irrelevant
`
`that it also “applies new user rules to abstractions”).
`
`F.
`
`“Video Device”
`
`Avigilon argues that the phrase “video device” in the preambles of claims 9
`
`and 30 is limiting. Paper 27, 14-15. But Avigilon does not state how the phrase
`
`should be construed. Moreover, as discussed below, Dr. Bovik admits Kellogg
`
`discloses a video device. Thus, this term need not be construed.
`
`IV. THE CHALLENGED CLAIMS ARE UNPATENTABLE
`
`A.
`
`Kellogg Discloses the “Single Camera” Limitations
`
`Avigilon asserts that Kellogg does not disclose “detecting an object in a
`
`video from a single camera” or “detecting a plurality of attributes of the object by
`
`analyzing the video from said single camera” because it says Kellogg discloses
`
`“multiple cameras.” Paper 27, 15-22. But this assertion is factually wrong and
`
`misreads the claims.
`
`First, the claims do not preclude systems with multiple cameras. Such an
`
`interpretation would directly contradict the ’923 patent specification, which
`
`consistently discloses the use of multiple cameras/image sensors. See, e.g., Ex.
`
`1001, 6:3-12 (“video sensors 14,” …examples of which include “a video camera”);
`
`9:23-24; 12:51-57; Fig. 1.
`
`10
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Avigilon does not cite any contrary disclosure in the ’923 patent because
`
`there is none. Nothing in the ’923 patent suggests that its inventors invented
`
`technology for single camera object detection. See Ex. 1056, 21:19-22:4 (Bovik
`
`admitting “object detection from a single camera” not “the invention”), 175:21-
`
`176:13 (patent does not teach how to solve any single camera object detection
`
`problems).
`
`It is nonsensical for Avigilon to argue that Kellogg is deficient because it
`
`discloses multiple cameras or that to be viable prior art Kellogg needed to contain
`
`detailed disclosure of how to use of a single camera when the ’923 patent itself
`
`does not do so.
`
`Moreover, Avigilon ignores that Kellogg does explicitly state that it
`
`performs the “real-time processing of CCD camera images.” Ex. 1003, 77. This is
`
`more direct support for single camera object detection than the ’923 patent.
`
`Kellogg further shows a GUI with a single camera in Figure 4-9, which Avigilon
`
`disregards because the GUI contains a scroll bar. Avigilon argues this disclosure is
`
`deficient because there might be other cameras off-screen. Paper 27, 15-19. But,
`
`the claim does not preclude other cameras. See Ex. 1056, 183:4-15 (Bovik stating
`
`“if there’s another camera doing something unrelated somewhere else, it’s
`
`completely unrelated to this claim”). It simply requires at least one camera to
`
`detect an object and the objects attributes, which is what Figure 4-9 shows.
`
`11
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`And, nothing in Kellogg suggests that video from the multiple cameras must
`
`be combined to perform object detection, as was the case with the prior art
`
`Carlbom reference distinguished during prosecution.
`
`Avigilon also alleges that Kellogg cannot disclose single-camera object
`
`detection because it does not describe any procedure for object detection at all.
`
`Paper 27, 20-21. But, Avigilon’s own expert, Dr. Bovik, admits that Kellogg
`
`clearly shows object detection and tracking:
`
`[T]here is no place in Kellogg that says that it is struggling with tracking
`objects or that objects were being occluded. Nor do I, or would any
`POSITA, view any aspect of Kellogg as indicating that it was a problem for
`the Kellogg reference.
`
`Ex. 2019, 29-30.
`
`
`
`This admission is unsurprising because the ’923 patent does not purport to
`
`have invented object tracking or detection, and instead states that any known
`
`method can be used. Ex. 1001, 9:34-35; 9:39-41. The detection and tracking
`
`algorithms identified by Kellogg and cited in the Petition certainly fit the
`
`specification’s declaration that “any” algorithm can be used. See Petition, 35-36,
`
`50. Indeed, the Board also confirmed this in the related ’661 IPR. See IPR2018-
`
`00138 FWD, 21 (“Kellogg specifically discloses detecting objects”).
`
`12
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Avigilon’s arguments that Kellogg does not disclose detecting a “plurality of
`
`attributes” from a single camera merely rely on its general single camera
`
`arguments. Paper 27, 21-22. Avigilon fails to submit any support. And, indeed,
`
`Kellogg of this feature is at least as fulsome as the corresponding disclosure in the
`
`’923 patent. See Petition, 37 (referencing, e.g., “position,” “height,” “person,” and
`
`“time” attributes).
`
`B.
`
`Kellogg Discloses “Selecting a New User Rule After Detecting the
`Plurality of Attributes”
`
`Avigilon criticizes the Petition for “focusing on the timing of when the new
`
`user rule is selected” (Paper 27, 22). But this is exactly what this limitation
`
`requires, selecting a new user rule “after” detecting attributes.
`
`Avigilon asserts without support that Kellogg does not detect attributes at
`
`all.4 Id. Yet, on the same page, Avigilon acknowledges that Kellogg searches the
`
`pre-collected centroid trajectories of moving objects. Id. And the ’923 patent
`
`explicitly teaches that an object’s trajectory is an attribute. Ex. 1001, 7:6-11 (a
`
`salient motion is a video primitive, i.e., attribute); 7:42-50 (trajectories are a
`
`property of a salient motion).
`
`4 Avigilon cites Section IV(A)(2) of its Response for this proposition, but nothing
`their addresses Kellogg’s attribute detection.
`
`13
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`Dr. Grindon did not admit that Kellogg queries detected events. Paper 27,
`
`22-23. To the contrary, Dr. Grindon directly rejected the assertion that Kellogg
`
`searches events:
`
`[Q] do you agree that the approach query, in searching trajectories of
`moving objects, is searching what the specification would call events?
`
`A. So you’re talking about an approach query of two objects? So that would
`have both physical and temporal attributes associated with the objects,
`which would be multiple attributes.
`
`Ex. 2018, 57:10-19.
`
`C.
`
`Kellogg Discloses “Applying the New User Rule to the Plurality of
`Detected Attributes”
`
`Avigilon argues that Kellogg stores and queries events (not attributes) like
`
`Courtney. Paper 27, 23-26. The Board has already addressed and rejected this
`
`argument, holding “Kellogg is distinguished from Courtney … Kellogg discloses
`
`the detection of attributes is not affected by the user rule.” See IPR2018-00138
`
`FWD, 20. Because the issue of the scope and content of Kellogg as compared to
`
`Courtney has already been actually litigated and rejecting Avigilon’s position was
`
`a necessary component of the Board’s final judgment in the ’661 IPR, Avigilon is
`
`collaterally estopped from relitigating this issue.
`
`Avigilon presents nothing new here compared to the ’661 IPR. Avigilon
`
`criticizes Kellogg because it stores motion attributes, including trajectories, in an
`
`14
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`object oriented database. Paper 27, 24. But, the ‘923 patent also detects objects
`
`and their attributes, including motion attributes. Ex. 1001, 7:1-12 (video primitives
`
`include velocity, speed and motion); 7:42-50.
`
`Nothing in the ’923 patent limits how motion attributes are stored–it is
`
`silent. Dr. Bovik agrees. See, e.g., Ex. 1056, 45:5-14 (there must be “some kind of
`
`storage involved,” but “it’s not claimed”), 55:24-56:4 (no opinion on whether
`
`storing objects in an index is within the claims), 85:15-23 (“not excluding” storing
`
`attributes “in an object-oriented database”), 114:15-18 (“patent doesn’t say” you
`
`cannot store “an object, its classification and its motion”). Logic dictates that the
`
`’923 patent system must record some connection between the object and motion
`
`attribute or it couldn’t otherwise retrieve that information later.
`
`Avigilon criticizes Kellogg for including details about how detected objects
`
`and attributes are stored that are absent in the ’923 patent specification. In reality,
`
`Kellogg discloses exactly what the claim requires—“applying the new user rule to
`
`the attributes”—regardless of how the attributes are stored.
`
`Avigilon cites scattered use of the word “event” in Kellogg (Paper 27, 26).
`
`But Avigilon’s expert admitted that the word event is used differently in Kellogg
`
`than the ’923 patent. See Ex. 1056, 119:21-24 (Bovik stating Kellogg is “not
`
`storing the same events as in the patent”). Kellogg operates just like the ’923
`
`patent. Kellogg can “analyze data provided by various cameras and store
`
`15
`
`
`
`IPR2019-00311
`U.S. Patent No. 7,932,923
`
`information in the visual memory,” “then retrieve this data to track objects, watch
`
`for suspicious events, and respond to user queries”. See, e.g., Ex. 1003, 10. This is
`
`what the ’923 patent discloses. See, e.g., Ex. 1001, 11:33-35 (“As the system
`
`operates, information regarding event occurrences is collected, and the
`
`information can be viewed by the operator at any time”).
`
`Dr. Grindon did not agree with Avigilon’s assertion that Kellogg does not
`
`store attributes. Paper 27, 24. He explicitly stated that “Kellogg does disclose
`
`storing of attributes.” Ex. 2018, 18:24-19:6. Dr. Grindon also did not
`
`“acknowledge[] that Kellogg stores and queries object-oriented data like in the
`
`Day-I reference.” Paper 27, 25.
`
`D.
`
`Kellogg Discloses “Applying the New User Rule to Only the
`Plurality of Detected Attributes”
`
`Avigilon’s argument that Kellogg does not apply its rules to only the
`
`attributes (Paper 27, 27-31), relating to Kellogg’s “fixed grid” and “PR quadtree”
`
`indices, is irrelevant because Petitioner does not rely on these features of Kellogg.
`
`See Petition, 40; Ex. 1005, ¶¶221-222. As Kellogg explains, “[i]n addition to” the
`
`fixed grid and PR quadtree indices, Kellogg “also includes a ‘bucket’ index,”
`
`which “simply maintains a list of all the objects stored in the visual memory” and
`
`“answers a query by retrieving all the objects in its list and checking them against
`
`the query specification.” Ex. 1003, 83.