Flock's patents

A granted patent is the one document where a company must describe truthfully what its technology does. These are Flock Group Inc.'s.

• US Patent 10,559,200 B1 — capturing significant surveillance images (2020-02-11)
  On-device neural-net (RCNN/YOLO) object identification; spec defines an 'identifiable object' to include 'a human, an animal, a vehicle'. Granted claims plate/vehicle-centered; human ID in the spec. Moderate weight.
• US Patent 10,795,933 B1 — object based query of video content (dynamic surveillance network) (2020-10-06)
  PARENT of US11416545B1. Shared spec: classifies people by 'male, female, race', clothing, height/weight; matches 'face recognition data points'; cross-camera person + vehicle path tracking; flags a specific named individual. Quote: 'classify the identified object ... the object is a human, is male, is wearing a jacket, and is about 5'11" and 190 pounds'. Strongest single contradiction.
• US Patent 11,416,545 B1 — System and method for object based query of video content captured by a dynamic surveillance network (2022-08-16)
  Patent describes querying surveillance VIDEO by OBJECT, not just plates: a neural-network module classifies people by 'male, female, race' and further aspects incl. clothing type and height/weight (Fig 5A), with per-object confidence scores. Directly contradicts Flock's public line that its tech 'cannot recognize, identify, or track individuals.' PDF stored locally, 29pp.
• US Patent 12,322,186 B1 — significant surveillance images via plurality of i/r wavelengths (2025-06-03)
  Dual-wavelength IR camera; detector 'trained to detect and localize license plates ... as well as other objects, such as vehicles ... and/or people'. Confirms 'person' object class still active 2025. Moderate weight.

Flock Patent Exhibit List

Purpose: Documentary evidence that Flock's public claim — that its technology "cannot recognize, identify, or track individuals" — is contradicted by Flock's own granted patents.

Anchor (context, already documented — not re-derived here): US11416545B1 classifies detected people by "male, female, race," clothing type, and height/weight. It is a Continuation of application No. 15/968,402 (now US10795933B1) and shares that patent's specification verbatim. The exhibits below are the independently verified patents that establish the same capabilities.

1. Exhibit List — Confirmed Contradictory Flock Patents

Exhibit A — US10795933B1 (the strongest single contradiction; direct legal ancestor of the anchor)

• Patent: US10795933B1 — System and method for object based query of video content captured by a dynamic surveillance network
• URL: https://patents.google.com/patent/US10795933B1/en
• Date: Granted 2020-10-06
• Assignee: Flock Group Inc
• Capability: This is the parent patent of the flagged anchor US11416545B1 (the anchor is expressly "a Continuation of application No. 15/968,402 ... now Pat. No. 10,795,933") and shares the identical specification. The shared spec claims classifying detected people by "people (male, female, race, etc.)", "clothing type," and "height/weight"; using "face recognition data points" to query the surveillance network for "a person with matching face recognition data points"; reconstructing a vehicle's path across many independent cameras to build a movement timeline; and identifying a specific known individual (an example describes a detected person flagged as "statistically similar to the teenage daughter of a subscriber").
• Verbatim contradiction quote:

> "the method applies one or more neural network algorithms to classify the identified object and aspects of the identified object (e.g., the object is a human, is male, is wearing a jacket, and is about 5′11″ and 190 pounds)"

• Why it contradicts: Directly contradicts Flock's claim that its technology "cannot recognize, identify, or track individuals" and is "not facial recognition." The specification recites biometric person classification (sex/race/height/weight/clothing), explicit "face recognition data points" matching, cross-camera person tracking, cross-camera vehicle path tracking, and identification of a specific named individual — the strongest single contradiction found, and the direct legal ancestor of the anchor.

Exhibit B — US10559200B1

• Patent: US10559200B1 — Method and system for capturing and storing significant surveillance images
• URL: https://patents.google.com/patent/US10559200B1/en
• Date: Granted 2020-02-11
• Assignee: Flock Group Inc.
• Capability: Original grant of the "significant surveillance images" family (parent of US11030892B1). On-camera pipeline that detects object motion, generates images, and applies neural-network (RCNN/YOLO) filters that identify objects and their aspects. The family specification defines an "identifiable object" as "any object ... whether it be a human, an animal, [or an] automobile" and describes an "algorithmic system customized to identify objects ... in a manner modeled after human cognitive processing."
• Verbatim contradiction quote:

> "Individual image frames contain \"content\" which is whatever was captured by the image whether it be a human, an animal, a vehicle, a tree, an empty parking lot, etc. In this description, an \"identifiable\" object is any object within a video frame or frame composite for which all relevant features are detected"

• Why it contradicts: Establishes on-device neural-network object identification whose disclosed scope explicitly includes humans, not merely license plates/vehicles — inconsistent with the public claim that the technology cannot recognize or identify individuals. Moderate weight: the granted claims center on vehicles/plates while the human-identification capability is disclosed in the specification.

Exhibit C — US12322186B1

• Patent: US12322186B1 — Method and system for capturing and storing significant surveillance images using a plurality of i/r wavelengths
• URL: https://patents.google.com/patent/US12322186B1/en
• Date: Granted 2025-06-03
• Assignee: Flock Group Inc
• Capability: Dual-wavelength infrared illumination camera for low-light/backlit surveillance capture with power conservation. The specification states the on-device object detector "identif[ies] each potential object (i.e. vehicle 120, person 125, and/or plate 115) within a captured image," using a neural network "trained to detect and localize license plates ... as well as other objects, such as vehicles ... and/or people."
• Verbatim contradiction quote:

> "the ML model may be trained to detect and localize license plates 115 (as well as other objects, such as vehicles 120 and/or people 125)."

• Why it contradicts: The disclosed object detector explicitly detects and localizes people (a "person" class), contradicting the claim that Flock's cameras cannot recognize/identify/track individuals. Moderate weight: the granted claims are hardware/illumination-focused (object = license plate); person detection is in the specification and figures.

2. The Through-Line

Taken as a portfolio rather than as isolated documents, these patents describe a coherent, deliberately engineered capability that Flock's public messaging denies. The earliest grant (US10559200B1, 2020) already defines an "identifiable object" to expressly include a human. The parent/anchor pair (US10795933B1 → US11416545B1) then goes far beyond mere detection: it claims classifying people by sex, race, height, weight, and clothing, matching "face recognition data points," tracking a person and a vehicle across many independent cameras, and flagging a specific named individual. The most recent grant (US12322186B1, 2025) confirms the "person" object class is still an active, trained part of the on-device detector five years later. A company does not draft, prosecute, and maintain a multi-year patent family teaching biometric human classification, face-recognition-data matching, and cross-camera person tracking unless it intends to build and protect exactly that. The intent recorded in the patent record — the document class where a company must tell the truth about what its technology does to secure a legal monopoly on it — is squarely at odds with the marketing claim that the same technology "cannot recognize, identify, or track individuals."

3. Weak / Plate-Only Patents (checked and set aside)

• US11030892B1 — Method and system for capturing and storing significant surveillance images (https://patents.google.com/patent/US11030892B1/en) — plate/vehicle-focused; set aside.
• US12242288B1 — Autonomous drone mesh sensor deployment system (https://patents.google.com/patent/US12242288B1/en) — deployment/hardware-focused, not people-identifying; set aside.

Rejected (not Flock, or unverifiable): 0.