Prior art

To identify the most relevant prior art for US Patent 10,650,285, I will access the patent document from the USPTO database and review its cited references. Since I do not have direct access to the USPTO database for live querying, I will rely on the information provided in the original patent text, which often lists "Prior art keywords" and "Related Applications" but also includes a full "References Cited" section that lists other patents and publications considered by the examiner.

Based on the provided patent text, the "Related Applications" section indicates that US10650285 is a continuation-in-part of and claims priority to U.S. patent application Ser. No. 15/714,376, filed September 25, 2017, which itself claims priority to U.S. Provisional Patent Application Ser. No. 62/398,665, filed September 23, 2016. These are part of the same patent family and are therefore not considered prior art for anticipation under 35 U.S.C. § 102 against US10650285, as they share common inventorship and priority dates.

The patent text also explicitly incorporates by reference two publications that serve as background for deep learning models:

1. Krizhevksy et al. “ImageNet Classification with Deep Convolutional Neural Networks”, Advances in neural information processing systems. 2012.

   • Publication/Filing Date: 2012
   • Brief Description: This publication describes Alexnet, an example of a Convolutional Neural Network (CNN) processing model used for image classification with deep learning.
   • Potential Anticipated Claims: This reference describes a foundational deep learning methodology (CNN) that is incorporated into the inventive process. The patent mentions that the machine learning classifier "includes a convolutional neural network (CNN) to preprocess the aerial image... and to classify the property features". Therefore, elements of using CNN for image processing and feature classification, as broadly described in this paper, could potentially anticipate aspects of claims that do not include the specific application to property characteristics from aerial imagery or the subsequent condition and risk analysis. For example, any claim element generally describing using a CNN for image processing or feature classification, without the specific context of property analysis, might be anticipated.

2. Lin et al. “Network In Network”, International Conference on Learning Representations, 2014 (arXiv:1409.1556).

   • Publication/Filing Date: 2014
   • Brief Description: This paper describes the Network in Network (NIN) model, which is a deep learning model that generates artificial perception outcomes using micro neural networks. The patent highlights NIN's superior performance and reduced storage intensity compared to conventional CNN processing.
   • Potential Anticipated Claims: Similar to the Krizhevksy et al. paper, this reference details a deep learning model (NIN) that the patent states "has demonstrated superior performance outcome to conventional CNN processing" and "is less storage-intensive than CNN processing." Claim elements generally describing the use of NIN for deep learning analysis could be anticipated. The patent explicitly states that the "deep learning analysis model may be NIN." Therefore, any claims broadly covering the application of NIN for image analysis or feature learning, without the specific context of property characteristic and condition assessment from aerial imagery, could potentially be anticipated.

It's important to note that these references are cited to provide background on deep learning methodologies. Anticipation under 35 U.S.C. § 102 requires that all elements of a patent claim be disclosed, either explicitly or inherently, in a single prior art reference. While these references describe aspects of deep learning, they do not, on their own, appear to describe the entire inventive method or system of US10650285, which involves the specific application of these techniques to aerial imagery for property characteristic classification, condition analysis, and subsequent risk/cost estimation. Therefore, they are more likely relevant for an obviousness analysis under 35 U.S.C. § 103, in combination with other references, than for anticipation.

The patent mentions "Prior art keywords" such as "property", "characteristic", "condition", "image", and "processing circuitry", but these are general terms and do not refer to specific prior art documents. The legal status section of Google Patents for US10650285B1 also indicates prior art keywords. The PTAB challenge, IPR2025-01357, cited U.S. Patent Application Publication No. 2017/0316524 ("Okazaki"), U.S. Patent No. 9,412,239 ("Patil"), and U.S. Patent No. 9,208,610 ("Yagnik") as prior art for obviousness grounds (35 U.S.C. § 103), not necessarily for anticipation (35 U.S.C. § 102). Since the prompt specifically asked for anticipation under 35 U.S.C. § 102, these references are not elaborated here as prior art for anticipation based on the available information.