PTAB

IPR2025-01357

Zestyai Inc v. Aon Re Inc

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Automated Property Assessment Using Aerial Imagery and Machine Learning
  • Brief Description: The ’285 patent discloses systems and methods for automatically analyzing aerial imagery to classify property characteristics (e.g., roof type) and assess their condition (e.g., new or damaged). The invention uses a two-stage machine learning process to first identify a property feature and then to classify its condition, primarily for use in property risk evaluation for the insurance industry.

3. Grounds for Unpatentability

Ground 1: Obviousness over Gross - Claims 1-7 and 9-22 are obvious over Gross.

  • Prior Art Relied Upon: Gross (Application # 2015/0186953).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Gross, a single prior art reference, teaches a comprehensive property assessment system that renders all challenged claims obvious under 35 U.S.C. §103. Gross discloses training a "Classifier Engine" and using it to assess and rate properties for insurance purposes. Petitioner asserted that Gross teaches the key limitations of independent claims 1, 9, and 16 by disclosing a system that obtains aerial images of properties (via drones, satellites); extracts features characterized by a "distinct set of pixels, shapes, proportions, curves"; and uses its Classifier Engine to determine both a characteristic classification (e.g., a roof's type is "pitched" or "shingle") and a condition classification (e.g., the roof is "new" or has "missing/damage tile"). Petitioner further argued that Gross’s disclosure of using neural networks and incorporating references that describe convolutional neural network (CNN) models renders the use of a CNN model, as recited in claim 5, obvious. Other dependent claims related to generating graphical reports, calculating repair costs, and assessing image orthogonality were also alleged to be disclosed or suggested by Gross.
    • Motivation to Combine (within Gross): The central argument for obviousness focused on the ’285 patent’s use of two separate machine learning classifiers. Petitioner contended that although Gross describes a single "Classifier Engine" performing both characteristic and condition classification, Gross also explicitly suggests that "it may be useful to employ multiple classifiers" and that classifiers can be "customized for a vendor... as corresponding to... elements, types and/or associated conditions." A POSITA would, therefore, find it an obvious and predictable design choice to implement Gross’s system using two specialized classifiers—one for characteristics and a second for conditions—to improve the system's accuracy and modularity.
    • Expectation of Success: A POSITA would have a high expectation of success in separating the classification tasks into two models. This approach represents a common and well-known engineering practice of dividing a complex problem into simpler, more manageable sub-problems to yield predictable improvements in performance and accuracy.

Ground 2: Obviousness over Gross and Davis - Claim 8 is obvious over Gross in view of Davis.

  • Prior Art Relied Upon: Gross (Application # 2015/0186953) and Davis (Patent 10,755,357).
  • Core Argument for this Ground:
    • Prior Art Mapping: This ground specifically targeted claim 8, which adds the limitation of "determining... a risk estimate of damage to the characteristic of the property due to one or more disasters." Petitioner argued that Gross teaches a general framework for insurance risk assessment based on observed property conditions (e.g., poor upkeep evidenced by "holes/cracks" in a roof) and other hazards like arson. Davis was asserted to supply the explicit teaching of assessing risk from specific disasters. Davis discloses a system for analyzing aerial images to determine a property's risk of damage from disasters like fire (based on roof construction type) or from a hurricane forecast. Petitioner argued that combining Gross's condition-based assessment with Davis's disaster-specific analysis meets all limitations of claim 8.
    • Motivation to Combine: A POSITA would combine the teachings of Gross and Davis to create a more robust and commercially valuable insurance risk assessment tool. Both references address the same problem (property risk assessment for insurance) using the same technical approach (image analysis from aerial photography). Combining Davis’s specific disaster modeling with Gross’s broader property condition assessment system would provide an insurer with more comprehensive and actionable information, which is the explicit goal of both references. The combination was presented as a simple aggregation of known elements to achieve a predictable result.
    • Expectation of Success: A POSITA would have readily been able to implement Davis's techniques in Gross's system with a reasonable expectation of success. Both systems use image analysis to evaluate property features for determining insurance risk. Integrating Davis’s rule-based disaster risk logic into Gross's machine learning framework was argued to be a straightforward application of known data correlation and system integration methods.

4. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-22 of Patent 10,650,285 as unpatentable.