Google LLC v. Cellular South Inc

1. Case Identification

• Case #: IPR2025-00877
• Patent #: 11,126,853
• Filed: May 10, 2025
• Petitioner(s): Google LLC
• Patent Owner(s): Cellular South, Inc.
• Challenged Claims: 1-11

2. Patent Overview

• Title: Video to Data
• Brief Description: The ’853 patent discloses a system for generating data from video content. The system uses a distributed architecture with a coordinator, splitter, and demultiplexer nodes to segment video, extract frames, and detect and recognize objects within those frames using fractal-based models.

3. Grounds for Unpatentability

Ground 1: Obviousness over Zhao, Kritt, Steinberg, and Kouzani - Claims 1-4 are obvious over Zhao in view of Kritt, Steinberg, and Kouzani.

• Prior Art Relied Upon: Zhao (Application # 2009/0141940), Kritt (Application # 2014/0181668), Steinberg (Patent 7,574,016), and Kouzani (a 1997 IEEE conference proceeding).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Zhao, the primary reference, discloses a core system for generating data from video by detecting, recognizing, and modeling objects like faces. However, Zhao lacks specific teachings for certain limitations of claim 1. Petitioner asserted that a person of ordinary skill in the art (POSITA) would have combined Zhao with other references to arrive at the claimed invention.
    • Kritt was cited to teach the distributed processing architecture recited in claim 1[a] ("coordinator," "splitter," "plurality of demultiplexer nodes"). Kritt described using scalable, cloud-based computing with distinct modules for video and audio processing, which Petitioner contended would have been a known and obvious way to implement Zhao’s system on a larger scale.
    • Steinberg was cited to teach the "adjustable" image detector of claim 1[b]. Steinberg disclosed a face detection process with an adjustable sensitivity or threshold, which allowed a user to either reduce false positives or increase the detection of less prominent (non-primary) faces, directly addressing the claimed adjustability.
    • Kouzani was cited to teach the fractal-based object recognizer of claim 1[c]. Kouzani described using fractal models for face recognition, touting benefits like computational efficiency and high data compression. Petitioner argued this was a known technique that a POSITA would have used to implement the object recognition step in Zhao’s system.
  • Motivation to Combine: A POSITA implementing Zhao’s video analysis system for a large-scale application (e.g., video indexing website) would have been motivated to improve its scalability, flexibility, and efficiency. Kritt provided a known solution for scalability through cloud computing. Steinberg offered a known method to improve the utility of object detection by making it adjustable. Kouzani provided a known, computationally efficient alternative for facial recognition, which is a key feature of Zhao’s system.
  • Expectation of Success: Petitioner asserted a POSITA would have had a reasonable expectation of success because the combinations involved applying conventional, well-established techniques (distributed processing, adjustable detection thresholds, fractal-based recognition) to their known fields of use in a predictable manner.

Ground 2: Obviousness over Ground 1 Combination plus Yang and Romdhani - Claims 5-7 are obvious over the Ground 1 prior art in further view of Yang and Romdhani.

• Prior Art Relied Upon: The Ground 1 combination, plus Yang (a 2002 IEEE journal article) and Romdhani (a 2002 computer vision conference paper).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground adds Yang and Romdhani to the base combination to address dependent claims 5-7, which recite determining "skin textures" and using a "three-dimensional morphable model." Petitioner argued that using skin texture as a feature for facial recognition was a known technique to improve accuracy. Both Yang and Romdhani were cited as teaching the use of skin and face textures for object detection and recognition. Romdhani was further cited for expressly disclosing a "3D Morphable Model" to achieve state-of-the-art results in face identification, directly mapping to the limitations of claims 6 and 7.
  • Motivation to Combine: A POSITA seeking to improve the accuracy of the facial recognition in the base system of Ground 1 would have been motivated to investigate additional, more discriminative features beyond basic geometric landmarks. Yang and Romdhani showed that using skin texture and 3D models were known and effective methods to achieve superior results.

Ground 3: Obviousness over Ground 1 Combination plus Trivedi - Claims 8-9 are obvious over the Ground 1 prior art in further view of Trivedi.

• Prior Art Relied Upon: The Ground 1 combination, plus Trivedi (Application # 2006/0187305).

• Core Argument for this Ground:

  • Prior Art Mapping: This ground adds Trivedi to address dependent claims 8-9, which recite adding a "camera for capturing the video" and a processor configured to "transform the video into a rectilinear format." Petitioner noted that Zhao’s system is useful for video surveillance, which inherently implies a camera. Trivedi was cited as express confirmation, disclosing a surveillance system with a camera. Trivedi also explicitly taught transforming distorted, wide-angle video into a rectilinear format to improve viewing and processing, directly teaching the limitation of claim 9.
  • Motivation to Combine: For a surveillance application of the base system, adding a camera is a fundamental necessity. A POSITA would have further been motivated to incorporate Trivedi’s teachings on rectilinear transformation to improve the quality of the video input, thereby enhancing the accuracy of the downstream object recognition process.

• Additional Grounds: Petitioner asserted an additional obviousness challenge for claims 10-11 over the Ground 1 prior art in further view of Singer (Application # 2011/0305394). This ground relied on Singer’s teachings of embedding timed metadata, including object coordinate locations and timestamps, directly into a video stream.

4. Arguments Regarding Discretionary Denial

• Petitioner argued that discretionary denial under Fintiv would be inappropriate. The co-pending district court litigation was in a very early, pre-answer stage with no trial date set, and Petitioner intended to move to stay the litigation pending the outcome of the IPR.
• Petitioner also argued that denial under §325(d) was not warranted. The primary references for the core grounds (Zhao, Kritt, Steinberg, Kouzani, Yang, Romdhani) were not considered during prosecution. While Trivedi and Singer were cited by the examiner, Petitioner argued it was relying on them for the same undisputed limitations, presenting them in a new combination not previously evaluated.

5. Relief Requested

• Petitioner requests institution of an inter partes review and cancellation of claims 1-11 of the ’853 patent as unpatentable.