`Tel: 571-272-7822
`
`
`Paper 6
`Date: March 9, 2020
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`ROBERT BOSCH LLC,
`Petitioner,
`v.
`MONUMENT PEAK VENTURES, LLC,
`Patent Owner.
`____________
`
`IPR2019-01473
`Patent 6,654,507 B2
`____________
`
`
`Before JONI Y. CHANG, MICHAEL R. ZECHER, and
`JULIET MITCHELL DIRBA, Administrative Patent Judges.
`
`DIRBA, Administrative Patent Judge.
`
`
`
`
`DECISION
`Granting Institution of Inter Partes Review
`35 U.S.C. § 314
`
`
`
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`
`I. INTRODUCTION
`Robert Bosch LLC (“Petitioner”) filed a Petition seeking institution of
`inter partes review of claims 1, 8, and 14 of U.S. Patent No. 6,654,507 B2
`(Ex. 1001, “the ’507 patent”). Paper 1 (“Pet.”). Monument Peak Ventures,
`LLC (“Patent Owner”) filed a Waiver of Preliminary Response. Paper 5.
`To institute an inter partes review, we must determine that the
`information presented in the Petition “shows that there is a reasonable
`likelihood that the petitioner would prevail with respect to at least 1 of the
`claims challenged in the petition.” 35 U.S.C. § 314(a). Based on the
`information presented in the Petition and the supporting evidence, we
`determine that Petitioner has demonstrated a reasonable likelihood that it
`would prevail in establishing the unpatentability of at least one of claims 1,
`8, and 14. Accordingly, we institute an inter partes review of these
`challenged claims.
`
`Related Matters
`A.
`The parties identify the following related proceeding as previously
`pending in district court: Monument Peak Ventures, LLC v. Bosch Security
`Systems, Inc., 1:18-cv-01335 (D. Del.). Pet. 37; Paper 3, 2 (Mandatory
`Notice). According to Petitioner, Patent Owner voluntarily dismissed this
`action without prejudice. Pet. 37. Petitioner identifies no other related
`matters. Id.
`Patent Owner states that U.S. Patent No. 7,092,573 (which is the
`subject of IPR2019-01020) includes claim terms that may be relevant to
`claims of the ’507 patent. Paper 3, 2. The Board denied institution of this
`proceeding. General Electric Co. v. Monument Peak Ventures, LLC,
`
`2
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`IPR2019-01020, Paper 7 (PTAB Sept. 23, 2019) (Decision Denying
`Institution).
`In addition to the matters identified by the parties, we identify the
`following other matters that may be related. See 37 C.F.R. § 42.8(b)(2).
`The ’507 patent incorporates by reference two patent applications that each
`list, as the first named inventor, the sole inventor of the ’507 patent.
`Ex. 1001, 1:7–15, code (75). These applications issued as U.S. Patent
`Nos. 6,282,317 and 6,654,506, and Petitioner has filed inter partes review
`petitions seeking review of these patents (in IPR2019-01472 and IPR2019-
`01474, respectively). Today, the Board enters decisions granting institution
`of both. Robert Bosch LLC v. Monument Peak Ventures, LLC, IPR2019-
`01472, Paper 6 (PTAB March 9, 2020); Robert Bosch LLC v. Monument
`Peak Ventures, LLC, IPR2019-01474, Paper 6 (PTAB March 9, 2020).
`Also, Petitioner filed inter partes review petitions relating to other
`patents owned by Patent Owner: U.S. Patent No. 7,035,461 (IPR2019-
`01475) and U.S. Patent No. 7,148,908 (IPR2019-01476). A decision
`whether to institute has not been entered in either of these proceedings.
`
`3
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`The Petition’s Asserted Ground
`B.
`Petitioner asserts the following ground of unpatentability:
`
`Claims Challenged 35 U.S.C. §
`1, 8, 14
`103(a)1
`
`References
`Toyama,2 Itti,3 Neubauer4
`
`Petitioner also relies on the testimony of Dr. John R. Grindon, D.Sc.,
`to support its contentions. Ex. 1004.
`
`Summary of the ’507 Patent
`C.
`The ’507 patent is titled “Automatically Producing an Image of a
`Portion of a Photographic Image.” Ex. 1001, code (54). The application
`that led to the ’507 patent was filed on December 14, 2000. Id. at code (22).
`The Specification of the ’507 patent explains that existing techniques
`for automatically cropping an image “cannot deal with images with
`nonuniform background.” Ex. 1001, 2:3–51. But, by identifying “the main
`subject within the digital image” and using “the identified main subject . . .
`to automatically zoom and crop the image,” the Specification claims that
`“the present invention produces high-quality zoomed or cropped images
`
`1 The Leahy-Smith America Invents Act (“AIA”), Pub. L. No. 112-29, 125
`Stat. 284, 285–88 (2011), revised 35 U.S.C. § 103 effective March 16, 2013.
`Because the challenged patent was filed before March 16, 2013, we refer to
`the pre-AIA version of § 103.
`2 Toyama, US 6,792,135 B1, filed Oct. 29, 1999, issued Sept. 14, 2004
`(Ex. 1005).
`3 Itti et al., “A Model of Saliency-Based Visual Attention for Rapid Scene
`Analysis,” IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE
`INTELLIGENCE, vol. 20, no. 11, Nov. 1998 (Ex. 1006).
`4 Neubauer et al., US 6,553,131 B1, filed Sept. 15, 1999, issued Apr. 22,
`2003 (Ex. 1007).
`
`4
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`automatically, regardless whether the background is uniform or not.” Id. at
`3:54–61.
`“MSD” (or “main subject detection”) identifies which regions in an
`image are most likely to contain a main subject of the image and “provides a
`measure of saliency or relative importance” of the different regions of the
`image. Ex. 1001, 4:42–50, 5:29–33. “In particular, a large number of
`features are extracted . . . to represent a wide variety of visual saliency
`properties,” and these features are “input into a tunable, extensible
`probability network to generate a belief map containing a continuum of
`values.” Id. at 4:67–4.
`The output of MSD used by the invention is a list of
`segmented regions ranked in descending order of their likelihood
`(or belief) as potential main subjects for a generic or specific
`application. This list can be readily converted into a map in
`which the brightness of a region is proportional to the main
`subject belief of the region. Therefore, this map can be called a
`main subject “belief” map. This “belief” map is more than a
`binary map that only indicates location of the determined main
`subject. The associated likelihood is also attached to each region
`so that regions with large values correspond to regions with high
`confidence or belief of being part of the main subject.
`Ex. 1001, 5:17–28 (emphasis added). The belief map may then be clustered
`into belief levels to quantize MSD beliefs and “reduce the (unnecessary)
`variation in the belief map.” Id. at 9:17–10:18; see id. at 7:16–22 (disclosing
`a process of clustering regions in a belief map). For example, “three levels
`allow for the main subject (high), the background (low), and an intermediate
`level (medium) to capture secondary subjects, or uncertainty, or salient
`regions of background.” Id. at 9:27–32.
`After creating the belief map, the ’507 patent “determines a zoom
`factor (e.g. 1.5x) and a crop window” for the image. Ex. 1001, 7:1–4; see id.
`
`5
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`at Fig. 3 (steps 200–203), Fig. 4, 7:40–42. “This zoom factor is selected by
`an automatic method based directly on the main subject belief map (e.g., an
`estimate of the size of the main subject).” Id. at 7:4–7. The crop window is
`initially centered at the “center-of-mass” of the beliefs (excluding beliefs
`that correspond to the background), but is “moved so that the entire crop
`window is within the original image.” Ex. 1001, 7:23–25, 8:5–9. The
`window may again be moved to include “all the regions of the highest belief
`values (‘main subject’)” (id. at 8:9–12) and to maximize “the sum of the
`belief values,” which facilitates inclusion of secondary subjects. Id. at 8:20–
`57; see id. at Figs. 7–10 (illustrating movement of crop window 80).
`The ’507 patent “crop[s] the image according to” the final position of
`the crop window. Ex. 1001, Fig. 3 (items 209, 211) (emphasis removed).
`
`D. Challenged Claims
`The Petition challenges claims 1, 8, and 14 of the ’507 patent.
`Claim 1 is independent, and claims 8 and 14 directly depend from claim 1.
`Independent claim 1 is reproduced below:
`1. A method of producing an image of at least a portion
`of a digital image, comprising:
`a) providing a digital image having pixels;
`b) computing a belief map of the digital image by using
`the pixels of the digital image to determine a series of features
`and using such features to assign a probability of a location of a
`main subject of the digital image in the belief map;
`c) determining a crop window having a shape factor and a
`zoom factor, the shape and the zoom factors determining a size
`of the crop window; and
`d) cropping the digital image to include a portion of the
`image of high subject content in response to the belief map and
`the crop window.
`
`6
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`Ex. 1001, 11:57–12:3.
`
`II. ANALYSIS
`The Level of Ordinary Skill in the Art
`A.
`Petitioner asserts that the level of ordinary skill in the art corresponds
`to “a bachelor’s degree in electrical engineering, computer science, physics,
`or a related field, and two to four years of experience (or the academic
`equivalent) in the field of computer/machine vision or image processing.
`Less work experience may be compensated by a higher level of education,
`and vice versa.” Pet. 10 (citing Ex. 1004 ¶ 17).
`For purposes of this Decision, we adopt the level of ordinary skill as
`articulated by Petitioner because, based on the current record, this proposal
`appears to be consistent with the ’507 patent, the asserted prior art, and
`supported by the testimony of Dr. Grindon.
`
`Claim Construction
`B.
`Because the Petition was filed after November 13, 2018, we interpret
`claim terms using “the same claim construction standard that would be used
`to construe the claim in a civil action under 35 U.S.C. 282(b).” 37 C.F.R.
`§ 42.100(b) (2019).5 Under this standard, the words of a claim generally
`are given their “ordinary and customary meaning,” which is the meaning the
`term would have to a person of ordinary skill at the time of the invention, in
`
`5 On October 11, 2018, the U.S. Patent and Trademark Office revised its
`rules to harmonize the Board’s claim construction standard with that used in
`federal district court. Changes to the Claim Construction Standard for
`Interpreting Claims in Trial Proceedings Before the Patent Trial and Appeal
`Board, 83 Fed. Reg. 51,340 (Oct. 11, 2018). This rule change applies to
`petitions filed on or after November 13, 2018. Id.
`
`7
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`the context of the entire patent including the specification. Phillips v. AWH
`Corp., 415 F.3d 1303, 1312–13 (Fed. Cir. 2005) (en banc).
`In light of Petitioner’s arguments and supporting evidence, the only
`claim term requiring construction is the claim term “belief map” because it
`appears to be a claim term coined by the ’507 patent. We determine that no
`other terms need to be construed for our determination of whether to
`institute a review. See Nidec Motor Corp. v. Zhongshan Broad Ocean
`Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017) (“[W]e need only construe
`terms ‘that are in controversy, and only to the extent necessary to resolve the
`controversy.’” (quoting Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200
`F.3d 795, 803 (Fed. Cir. 1999))).
`
`“belief map”
`Claim 1 recites, in relevant part, “computing a belief map of the
`digital image by using the pixels of the digital image to determine a series of
`features and using such features to assign a probability of a location of a
`main subject of the digital image in the belief map.” Ex. 1001, 11:60–64.
`Petitioner proposes to construe a “belief map” as “a map indicating
`importance of a region by providing a measure of relative importance for
`different regions in an image.” Pet. 11–12. Petitioner’s declarant, Dr.
`Grindon, testifies that “[s]ynonyms for ‘belief’ maps are ‘likelihood,’
`‘probability,’ or ‘saliency’ maps.” Ex. 1004 ¶ 34. Patent Owner did not file
`a Preliminary Response. Paper 5.
`As noted above, the Specification discloses:
`The output of [main subject detection] used by the
`invention is a list of segmented regions ranked in descending
`order of their likelihood (or belief) as potential main subjects for
`a generic or specific application. This list can be readily
`
`8
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`converted into a map in which the brightness of a region is
`proportional to the main subject belief of the region. Therefore,
`this map can be called a main subject “belief” map. This
`“belief” map is more than a binary map that only indicates
`location of the determined main subject. The associated
`likelihood is also attached to each region so that regions with
`large values correspond to regions with high confidence or belief
`of being part of the main subject.
`Ex. 1001, 5:17–28 (emphases added). The Specification also discloses that
`“secondary main subjects are indicated by lower belief values in the main
`subject belief map, and can be included according to a descending order of
`belief values once the main subject of highest belief values are included.”
`Id. at 5:51–54.
`Based on the evidence in this present record, we agree with the
`Petitioner’s proposed claim construction, and we interpret the term “belief
`map” as “a map indicating importance of a region by providing a measure of
`relative importance for different regions in an image,” for purposes of this
`Decision, because the proposed construction is consistent with the claim
`language and Specification.
`
`Principles of Law on Obviousness
`C.
`The legal question of obviousness is resolved on the basis of
`underlying factual determinations including (1) the scope and content of the
`prior art; (2) any differences between the claimed subject matter and the
`prior art; (3) the level of ordinary skill in the art; and (4) when in evidence,
`objective evidence of non-obviousness.6 Graham v. John Deere Co. of Kan.
`
`
`6 The current record does not include allegations or evidence of objective
`indicia of non-obviousness.
`
`9
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`City, 383 U.S. 1, 17–18 (1966). One seeking to establish obviousness based
`on more than one reference also must articulate sufficient reasoning with
`rational underpinnings to combine teachings. See KSR Int’l Co. v. Teleflex
`Inc., 550 U.S. 398, 418 (2007).
`
`D. Obviousness in view of Toyama, Itti, and Neubauer
`Petitioner contends that claims 1, 8, and 14 are rendered obvious by
`the combination of Toyama, Itti, and Neubauer. Pet. 13–37. Patent Owner
`did not file a Preliminary Response. Having considered the arguments and
`evidence presented in the Petition, we are persuaded at this stage of the
`proceeding that Petitioner has demonstrated a reasonable likelihood of
`prevailing in showing that at least one of claims 1, 8, and 14 would have
`been obvious over the combined teachings of Toyama, Itti, and Neubauer.
`
`1. Toyama (Ex. 1005)
`Toyama is titled “System and Method for Face Detection Through
`Geometric Distribution of a Non-Intensity Image Property.” Ex. 1005, code
`(54). Toyama describes a method for detecting a human face within an
`image, which is the first step in some computer vision applications, such as
`facial recognition and interpretation. Id. at code (57), 1:16–43.
`Toyama “determines a sub-region of the image to examine” and then
`performs detailed analysis on that sub-region to determine if a face is
`present. Ex. 1005, 6:25–32, Fig. 3. The Petition relies on Toyama’s
`identification of a sub-region (see generally Pet. 22–35); this is performed
`by hypothesis module 300, illustrated in Figure 4 and reproduced below:
`
`10
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`
`
`As shown in Figure 4, reproduced above, hypothesis module 300 includes
`generation module 400 and cropping module 410. Ex. 1005, 7:2–10.
`Generation module 400 “generates a hypothesis” and sends it to cropping
`module 410, which “then defines the dimensions and shape of a sub-region
`(or cropped image) based on the generated hypothesis.” Id. at 7:11–13,
`7:22–26; see id. at 7:4–7. Specifically, Toyama states:
`The generation module 400 receives raw image (box 420)
`and generates a hypothesis about the location of a face in the raw
`image (box 430). The hypothesis may include, for example,
`information about which image scales, aspect ratios and
`locations to examine. In a preferred embodiment of the
`invention, hypotheses are generated that include rectangular sub-
`regions of the image within a range of scales and at all possible
`
`11
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`image locations. Alternatively, hypothesis generation may
`include other types of vision processing that target regions of the
`image most likely to contain a face (such as regions of the image
`that contain skin color or ellipse-shaped blobs). The generated
`hypothesis is then sent as output (box 440) to the cropping
`module 410.
`The cropping module 410 then defines the dimensions and
`shape of a sub-region (or cropped image) based on the generated
`hypothesis (box 450). The dimensions and shape are applied to
`the raw image (box 460) and a cropped image is sent as output
`(box 470). . . .
`Ex. 1005, 7:11–28; see id. at 10:25–11:26 (providing “working example”
`where “sub-regions were defined for a limited range of scales” such that a
`small face in the background was not detected), 12:16–17 (claiming system
`where “the hypothesis is based at least partially on” an image property other
`than intensity). Later, a separate resizing module “resizes the cropped image
`to an optimal (or canonical) size” to facilitate Toyama’s detailed analysis.
`Id. at 7:62–64, Fig. 5 (item 500).
`
`12
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`2. Itti (Ex. 1006)
`Itti is an IEEE article7 titled “A Model of Saliency-Based Visual
`Attention for Rapid Scene Analysis.” Ex. 1006, 1254. Itti describes a
`framework for “fast selection of a small number of interesting image
`locations” using saliency maps. Id. at 1254; see id. at 1257 (“Although the
`concept of a saliency map has been widely used in [focus of attention]
`models . . . , little detail is usually provided about its construction and
`dynamics.”)
`After receiving an image, Itti generates various feature maps, which
`are combined into a “saliency map.” Ex. 1006, 1254–55. “In total, 42
`feature maps are computed: six for intensity, 12 for color, and 24 for
`orientation.” Id. at 1255. The “[f]eature maps are combined into three
`‘conspicuity maps’”—one for the intensity features, another for the color
`features, and a third for the orientation features8—and these conspicuity
`maps are “normalized and summed into the final input S to the saliency
`
`7 Petitioner contends that Itti qualifies as prior art. See Pet. 3, 17. In
`support of its assertion, Petitioner relies on the IEEE copyright line on the
`first page of the article, as well as printouts from the IEEE Xplore database
`regarding the magazine generally and this article specifically. Id. at 17
`(citing Exs. 1006, 1008, 1010). Patent Owner has not challenged the prior
`art status of this reference at this stage of the proceeding. On this record, we
`conclude that Petitioner has submitted evidence sufficient to establish a
`reasonable likelihood that Itti was publicly accessible in November 1998,
`and, thus, Petitioner has established a reasonable likelihood that the
`reference qualifies as prior art to the ’507 patent under 35 U.S.C. § 102(a)
`and (b).
`8 Itti explains, “The motivation for the creation of three separate channels”
`for the intensity, color, and orientation conspicuity maps “is the hypothesis
`that similar features compete strongly for saliency, while different
`modalities contribute independently to the saliency map.” Ex. 1006, 1255.
`
`13
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`map.” Id. “[T]he maximum of the saliency map (SM) defines the most
`salient image location, to which the focus of attention (FOA) should be
`directed.” Id.
`Itti’s Figure 3 includes an example input image and resulting saliency
`map (S), which is reproduced below, in relevant part.
`
`
`The portion of Figure 3 reproduced above shows, from top to bottom: an
`example input image; the color (C̄ ), intensity (Ī), and orientation (Ō)
`conspicuity maps for the image; and the resulting saliency map (S).
`Ex. 1006, 1255–56.
`
`3. Neubauer (Ex. 1007)
`Neubauer is titled “License Plate Recognition with an Intelligent
`Camera.” Ex. 1007, code (54). To recognize the characters on a license
`plate, Neubauer first performs a “coarse localization” on the image to
`
`14
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`identify the region within the image where the license plate is located. Id. at
`4:1–33, Fig. 1; see also id. at 3:46–57 (identifying benefits of using
`hierarchical classification and coarse segmentation methods). Neubauer’s
`coarse localization generates a “saliency map” that is used to determine the
`“region of interest.” Id. at 4:16–33, Fig. 1. Neubauer states:
`A saliency map blends vertical edges to achieve intensity regions
`in the image. Highest peak values in the intensity regions have
`the highest probability of being the license plate and are selected
`for further processing.
`Ex. 1007, 4:27–29.
`
`4. Independent Claim 1
`a. “A method of producing an image of at least a portion
`of a digital image, comprising”
`Petitioner asserts that Toyama discloses the preamble of claim 1.9
`Pet. 22–23. According to Petitioner, Toyama’s hypothesis module, which
`includes generation module 400 and cropping module 410, “generates a
`hypothesis and defines the dimensions of a sub-region, or cropped image, in
`a received raw image where a face may be found.” Id. at 23 (citing Ex.
`1005, 3:52–57, 6:32–36, Figs. 3, 4).
`
`b. “providing a digital image having pixels”
`Petitioner asserts that Toyama discloses this element. Pet. 23–24.
`Petitioner contends that raw image 200, which is input into Toyama’s face
`
`
`9 Petitioner treats the preamble of claim 1 as a limitation. Pet. 22–23. In
`this Decision, we do not decide that the preamble is limiting in this
`Decision; however, to the extent it is limiting, we are persuaded that
`Petitioner has sufficiently shown that it is disclosed by Toyama.
`
`15
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`detection system, maps to the claimed “digital image.” Id. at 23. According
`to Petitioner, Toyama discloses that this raw image is a digital image having
`pixels. Id. at 23–24 (citing Ex. 1005, 2:35–37, 7:2–10, 7:62–64, 8:9–15,
`8:21–37, 8:44–48, 10:37–39).
`
`c. “computing a belief map of the digital image by
`using the pixels of the digital image to determine a
`series of features and using such features to assign a
`probability of a location of a main subject of the
`digital image in the belief map”
`Petitioner asserts that the combination of Toyama, Itti, and Neubauer
`teaches or suggests this limitation. Pet. 24–32. In particular, Petitioner
`contends, “Toyama suggests but does not expressly disclose computing a
`belief map as claimed, and Itti and Neubauer resolves [sic] any dispute that
`computing a belief map and assigning a probability of a main subject’s
`location was known.” Id. at 24.
`First, Petitioner argues that Toyama suggests computing the claimed
`belief map. Pet. 25–27; see also id. at 30. Petitioner submits that Toyama’s
`generation module 400 provides a hypothesis to cropping module 410 “that
`identifies the regions ‘most likely’ to contain a face in the raw image.” Id. at
`25–26 (emphasis added) (quoting Ex. 1005, 7:13–20) (citing Ex. 1005,
`Fig. 4). Petitioner contends that this suggests that Toyama’s hypothesis is
`provided via a belief map that identifies the likely regions of interest. Id. at
`25, 27 (citing Ex. 1004 ¶¶ 53, 54, 56).
`In addition, Petitioner contends that, to the extent that Toyama does
`not disclose computing the claimed belief map, Itti discloses “computing a
`belief map by using the pixels to determine a series of features.” Pet. 27.
`Petitioner asserts that the claimed belief map is disclosed by Itti’s saliency
`
`16
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`map, which “defines the most salient image location to which the focus of
`attention (FOA) should be directed.” Id. (quoting Ex. 1006, 1255).
`According to Petitioner, Itti constructs feature maps from image pixels and
`combines these maps to create the saliency map. Id. at 27–28, 30 (citing
`Ex. 1006, 1254–55, Fig. 3).
`Petitioner contends that Toyama provides a motivation to combine the
`teachings of these references, as Toyama states that “the hypothesis
`generation may include ‘other types of vision processing that target regions
`of the image most likely to contain a face.’” Pet. 29 (emphasis added)
`(quoting Ex. 1005, 7:18–20). Petitioner asserts that, because Toyama
`“suggests a certain functionality but does not provide a detailed
`implementation of it,” a person of ordinary skill in the art would have looked
`to other references—such as Itti—“to fill [this] gap.” Id. at 25 (citing Ex.
`1004 ¶ 53).
`Petitioner also asserts that “inputting Itti’s saliency map into
`Toyama’s cropping module is nothing more than using a known technique to
`improve a similar system in the same way.” Pet. 29 (citing Ex. 1004 ¶ 61).
`Petitioner argues that Itti “teaches that its framework can be ‘easily tailored
`to arbitrary tasks through the implementation of dedicated feature maps.”
`Id. (quoting Ex. 1006, 1259). And, according to Petitioner, Toyama
`“discloses which features (skin color and ellipse-shaped blobs) should be
`used (or more heavily weighted) to specifically target faces.” Id. at 28
`(citing Ex. 1005, 7:18–23); see also id. at 26 (“Toyama further discloses that
`the hypothesis may be based ‘at least partially on the non-intensity image
`property.’” (quoting Ex. 1005, 12:16–17)). As a result, Petitioner contends
`that “combining Toyama and Itti would have been obvious because it was
`
`17
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`known to use saliency maps for image detection and Toyama teaches which
`features should be used.” Id. at 28 (citing Ex. 1004 ¶ 59); see also id. at 30
`(arguing the same (citing Ex. 1004 ¶ 62)).
`Further, Petitioner argues that Neubauer further demonstrates that a
`person of ordinary skill would have known how to implement a saliency
`map in this manner and there was a reasonable expectation of success in
`doing so. Pet. 28–29 (citing Ex. 1007, 4:26–30; Ex. 1004 ¶ 60). Petitioner
`also asserts both Toyama and Itti “involve detecting image features and
`focusing on the detected image region.” Id. at 29. Finally, according to
`Petitioner, Itti discloses several advantages of using its saliency map. Id. at
`29–30 (citing Ex. 1006, 1254, 1259).
`Moreover, Petitioner presents an alternative argument—Petitioner
`contends that, to the extent Toyama and Itti do not “assign a probability of a
`location of a main subject,” this limitation is disclosed by Neubauer.
`Pet. 31. Petitioner asserts, “Neubauer’s system generates a saliency map
`where the ‘highest peak values in the intensity regions have the highest
`probability of being the license plate and are selected for further
`processing.’” Id. (quoting Ex. 1007, 4:29–31 (emphasis added)) (citing
`Ex. 1007, 2:3–11, 5:4–7). According to Petitioner, “highest peak regions are
`selected for further processing because these peaks correspond to the highest
`probability, thus the system assigns a probability to the highest peak
`regions.” Id. (citing Ex. 1004 ¶ 63). Petitioner contends that it would have
`been obvious to a person of ordinary skill in the art to use Neubauer’s
`technique in Toyama’s facial recognition because Neubauer provides a
`computationally efficient solution to identify the relevant region of an
`image. Id. at 31–32 (citing Ex. 1007, 5:3; Ex. 1004 ¶ 64).
`
`18
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`d. “determining a crop window having a shape factor
`and a zoom factor, the shape and the zoom factors
`determining a size of the crop window” and
`“cropping the digital image to include a portion of
`the image of high subject content in response to the
`belief map and the crop window”
`Petitioner asserts that Toyama discloses these elements. Pet. 32–35.
`Petitioner submits that Toyama’s “cropping module [410] ‘defines the
`dimensions and shape of a sub-region (or cropped image) based on’ a
`previously generated hypothesis.” Id. at 32 (quoting Ex. 1005, 7:24–26)
`(citing Ex. 1005, Fig. 4, item 450). According to Petitioner, Toyama further
`discloses that its crop window has a zoom factor “because it discloses that
`sub-regions are ‘defined for a limited range of scales’ and that each cropped
`image is resized into a canonical image size.” Id. (quoting Ex. 1005, 10:46–
`49) (citing Ex. 1005, 10:53–55); see also id. at 34 (“By setting the sub-
`region to a specific scale, Toyama sets the zoom factor for the cropped
`images.” (citing Ex. 1004 ¶ 67)). Finally, Petitioner contends that Toyama’s
`cropping module 410 “applies a sub-region’s dimensions and shapes based
`on a generated hypothesis and outputs a cropped image” of the region most
`likely to contain a face. Id. at 34 (citing Ex. 1005, 7:18–22, 7:24–34).
`According to Petitioner, it would have been obvious to a person of ordinary
`skill in the art “to include the high subject content image portion in response
`to the belief map as disclosed by Itti because Itti’s framework can be ‘easily
`tailored to arbitrary tasks through the implementation of dedicated feature
`maps.’” Id. at 35 (quoting Ex. 1006, 1259)
`
`19
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`e. Conclusion
`Petitioner’s foregoing assertions and explanations are consistent with
`and supported by the evidence cited by Petitioner. In addition, Patent Owner
`has identified no deficiencies in Petitioner’s contentions at this stage of the
`proceeding. On this record, we are persuaded that Petitioner sufficiently has
`shown that Toyama discloses, or the combination Toyama, Itti, and
`Neubauer teaches or suggests, all the limitations of claim 1. Accordingly,
`we determine that Petitioner has demonstrated a reasonable likelihood of
`prevailing in its challenge to claim 1.
`
`5. Dependent Claims 8 and 14
`We also find that Petitioner has made an adequate showing that
`Toyama discloses the additional limitations of dependent claims 8 and 14.
`See Pet. 36–37 (citing Ex. 1005, 2:23–25, 4:15–27, 4:30–33, 7:28–30).
`Accordingly, we conclude that Petitioner also has demonstrated a reasonable
`likelihood of demonstrating that those claims would have been obvious over
`the combination of references.
`
`III. CONCLUSION
`For the foregoing reasons, we determine that Petitioner has
`established a reasonable likelihood of prevailing in its challenge to claims 1,
`8, and 14 of the ’507 patent. Accordingly, we institute an inter partes
`review of these claims on the asserted ground.
`Our determination in this Decision is not a final determination on
`either the patentability of any challenged claims or the construction of any
`claim term and, thus, leaves undecided any fact issues necessary to
`determine whether sufficient evidence supports Petitioner’s contentions by a
`
`20
`
`
`
`IPR2019-01473
`Patent 6,654,507 B2
`preponderance of the evidence in the final written decision. See Trivascular,
`Inc. v. Samuels, 812 F.3d 1056, 1068 (Fed. Cir. 2016) (noting that “there is a
`significant difference between a petitioner’s burden to establish a
`‘reasonable likelihood of success’ at institution, and actually proving
`invalidity by a preponderance of the evidence at trial” (quoting 35 U.S.C.
`§ 314(a) and comparing 35 U.S.C. § 316(e))).
`
`IV. ORDER
`
`It is:
`ORDERED that an inter partes review is instituted on all of the
`challenged claims, i.e., claims 1, 8, and 14 of the ’507 patent, on the sole
`ground of unpatentability specified in the Petition, which is identified in the
`Table in Section I.B. of this Decision; and
`FURTHER ORDERED that pursuant to 35 U.S.C. § 314(c) and
`37 C.F.R. § 42.4(b), inter partes review of the ’507 patent shall commence
`on the entry date of this Order, and notice is hereby given of the in