Arashi Vision Inc v. GoPro Inc

1. Case Identification

• Case #: IPR2024-01165
• Patent #: 10,958,840
• Filed: July 12, 2024
• Petitioner(s): Arashi Vision (U.S.) LLC (d/b/a Insta360)
• Patent Owner(s): GoPro, Inc.
• Challenged Claims: 1-21

2. Patent Overview

• Title: Video Stabilization
• Brief Description: The ’840 patent discloses systems and methods for stabilizing video content from an image capture device. The technology generates a smoothed motion trajectory based on position sensor data and a "look-ahead" analysis of future device movements, then uses this smoothed trajectory to define a cropped viewing window within the larger captured video frame to produce a stabilized output.

3. Grounds for Unpatentability

Ground 1: Obviousness over Zhou and Cai - Claims 1-3, 5-11, 14-16, 18, 19, and 21 are obvious over Zhou in view of Cai.

• Prior Art Relied Upon: Zhou (Patent 9,674,438) and Cai (Patent 10,158,802).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner asserted that Zhou, as the primary reference, taught the core elements of independent claim 1. Zhou described an image capture device with position sensors (gyroscopes, accelerometers) that generate motion data to determine a trajectory. It stabilized video by generating a smoothed trajectory based on both past and future motion data (a "look-ahead"), which is then used to place a smaller "visible frame" (the viewing window) within a larger "captured frame" to create the stabilized output. Petitioner argued Zhou also disclosed using parameters analogous to the claimed "low-light high-pass parameter" (by adjusting stabilization strength based on motion blur) and "stickiness parameter" (via a Gaussian weighting function that controls the impact of preceding trajectory portions).
  • Petitioner contended that to the extent Zhou did not explicitly teach using a "temporal horizon of motion" to approximate "intentional motion," Cai supplied this element. Cai taught a video stabilization method that distinguishes between "zero motion" and "intended motion" by analyzing a sequence of future frames (e.g., the next five frames) to characterize the movement more accurately.
  • Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine Cai with Zhou to improve Zhou's stabilization capabilities. Both references share the common goal of video stabilization. Petitioner argued that incorporating Cai's known technique for distinguishing intentional motion from jitter would have been an obvious way to enhance the accuracy and quality of the stabilization taught by Zhou.
  • Expectation of Success: A POSITA would have had a reasonable expectation of success in this combination, as it involved applying a known method (from Cai) to improve a similar system (Zhou's) to achieve a more robust and predictable outcome.

Ground 2: Obviousness over Zhou, Cai, and Karpenko - Claims 4 and 20 are obvious over the combination of Zhou, Cai, and Karpenko.

• Prior Art Relied Upon: Zhou (Patent 9,674,438), Cai (Patent 10,158,802), and Karpenko (Patent 9,071,756).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground built upon the Zhou and Cai combination to address the limitations of claim 4, which required the smoothed trajectory to be determined via "iterations of smoothing including a fine scale approach...and a coarse approach" that result in convergence. Petitioner argued Karpenko taught a constraint-based rotation smoothing algorithm and explicitly disclosed improving its computational efficiency by running it in a "coarse to fine mode." Karpenko stated this approach improves the "rate of convergence" and allows the algorithm to run with significantly fewer iterations.
  • Motivation to Combine: Petitioner asserted that since Zhou teaches computationally and power-efficient techniques, a POSITA seeking to optimize the Zhou/Cai system would be motivated to adopt Karpenko’s disclosed "coarse to fine" iterative smoothing method to improve the efficiency of calculating the smoothed trajectory.
  • Expectation of Success: Success would be expected because Karpenko's iterative smoothing is constraint-based, making it conceptually compatible with the system described in Zhou.

Ground 3: Obviousness over Zhou, Cai, and Sokeila - Claim 12 is obvious over the combination of Zhou, Cai, and Sokeila.

• Prior Art Relied Upon: Zhou (Patent 9,674,438), Cai (Patent 10,158,802), and Sokeila (Patent 8,994,838).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground addressed claim 12, which depends from claim 11 and adds the requirement that the buffer for storing visual content be a "circular buffer." While the primary combination taught using a buffer, it did not specify the type. Petitioner argued that Sokeila, another video stabilization reference, explicitly disclosed using a "ring buffer" (synonymous with a circular buffer) to hold video frames.
  • Motivation to Combine: A POSITA implementing the Zhou/Cai system would recognize the need to select a specific type of buffer. Petitioner argued it would have been an obvious design choice to implement a well-known buffer type, such as the circular buffer taught by Sokeila, from the finite, known options available.
  • Expectation of Success: Applying a known component (a circular buffer) to a known system to perform its conventional function (storing frames) would yield predictable results.
• Additional Grounds: Petitioner asserted an additional obviousness challenge against claims 13 and 17 based on the combination of Zhou, Cai, and the Bell Paper. This ground argued Bell Paper taught minimizing specific types of motion (rotational velocity and acceleration) to improve the general rotation stabilization taught by Zhou.

4. Arguments Regarding Discretionary Denial

• Petitioner argued that the Board should not exercise discretionary denial. It contended that denial under Fintiv is inappropriate because the Fintiv factors are directed at parallel district court litigation, not parallel U.S. International Trade Commission (ITC) proceedings, which cannot invalidate a patent.
• Petitioner also argued that denial under §325(d) is unwarranted because the petition raised new prior art and arguments not substantively considered during prosecution. Three of the five references (Cai, Sokeila, Bell Paper) were never before the Examiner. Further, the primary combination of Zhou and Cai was never considered by the Examiner, who instead focused on different prior art.

5. Relief Requested

• Petitioner requests the institution of an inter partes review and the cancellation of claims 1-21 of Patent 10,958,840 as unpatentable.