US-12620041-B1 - Infringement detection system

US12620041B1US 12620041 B1US12620041 B1US 12620041B1US-12620041-B1

Abstract

Concepts and technologies are disclosed herein for an infringement detection system that obtains images of products and images of proprietary objects, and analyzes the images to make coarse matches. An image comparison engine may transform or augment the data for comparison for coarse and refined matching. The outputs of the image comparison engine are initial infringement predictions that are further evaluated using refined matching including shape fitting. The detection system outputs refined infringement predictions, which may be optionally confirmed as counterfeit based on various considerations (e.g., known inauthentic products, suspect sales history, text and image anomalies, etc.). Upon a refined match or confirmation, the detection system records potential infringements and associated metadata into a database to initiate an optional response action. Multiple response actions are possible for transmission over a computer network to one or more receiving electronic addresses, computer servers, network system gateways or mobile/fixed devices.

Inventors

Barry Brager
Craig Meyer

Assignees

Barry Brager
Craig Meyer

Dates

Publication Date: 20260505
Application Date: 20231220

Claims (20)

1 . A system comprising: a noninvasive scanning device that comprises an image sensing device that is configured to capture images of a container and an emitter that is configured to emit energy toward the container; a processor that is coupled to the noninvasive scanning device; and a memory that stores computer-executable instructions that, when executed by the processor, cause the processor to perform operations comprising generating the images of the container using the noninvasive scanning device, wherein the images of the container comprise a representation of an object; obtaining, from an intellectual property data source, an intellectual property asset image that is included in an intellectual property asset; performing, using a neural network, a coarse match operation to identify a coarse match comprising an image pair, the image pair comprising one image of the images of the container and the intellectual property asset image; applying an image modification operation to one of the intellectual property asset image or the one image of the images of the container to obtain an augmented intellectual property asset image or an augmented image of the container associated with the object; creating, using the neural network and based on the coarse match, a refined match comprising one of: a first pairing of the augmented intellectual property asset image and the one image of the images of the container, a second pairing of the intellectual property asset image and the augmented image of the container, or a third pairing of the augmented intellectual property asset image and the augmented image of the container; and in response to creating the refined match, communicating predicted infringement data to a recipient device via an application programming interface.
2 . The system of claim 1 , wherein the computer-executable instructions, when executed by the processor, cause the processor to perform operations further comprising: identifying a domain of interest associated with the object, wherein the domain of interest comprises a type associated with the object; and determining, based on the domain of interest, an intellectual property classification associated with the type, the intellectual property classification comprising a class and a subclass, wherein the intellectual property asset is identified from a plurality of intellectual property assets in the class and the subclass.
3 . The system of claim 1 , wherein the intellectual property asset image comprises a patent drawing from a patent, a photograph from a patent, an image from a trademark registration, a trade dress drawing, or a mask work from a copyright registration.
4 . The system of claim 1 , wherein communicating the predicted infringement data to the recipient device comprises communicating the predicted infringement data to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement data.
5 . The system of claim 1 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice prompt to a chatbot.
6 . The system of claim 1 , wherein the neural network is trained using an independent loss function to learn intellectual property classification information, and wherein a portion of the intellectual property classification information and the intellectual property asset image are determined based on an intellectual property record.
7 . The system of claim 1 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice alert to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement notice alert.
8 . The system of claim 1 , wherein the noninvasive scanning device comprises a luggage scanner or a shipping container scanner.
9 . A device comprising: a noninvasive scanning device that comprises an image sensing device that is configured to capture images of a container and an emitter that is configured to emit energy toward the container; a processor that is coupled to the noninvasive scanning device; and a memory that stores computer-executable instructions that, when executed by the processor, cause the processor to perform operations comprising generating the images of the container using the noninvasive scanning device, wherein the images of the container comprise a representation of an object; obtaining, from an intellectual property data source, an intellectual property asset image that is included in an intellectual property asset; performing, using a neural network, a coarse match operation to identify a coarse match comprising an image pair, the image pair comprising one image of the images of the container and the intellectual property asset image; applying an image modification operation to one of the intellectual property asset image or the one image of the images of the container to obtain an augmented intellectual property asset image or an augmented image of the container associated with the object; creating, using the neural network and based on the coarse match, a refined match comprising one of: a first pairing of the augmented intellectual property asset image and the one image of the images of the container, a second pairing of the intellectual property asset image and the augmented image of the container, or a third pairing of the augmented intellectual property asset image and the augmented image of the container; and in response to creating the refined match, communicating predicted infringement data to a recipient device via an application programming interface.
10 . The device of claim 9 , wherein the intellectual property asset image comprises a patent drawing from a patent, a photograph from a patent, an image from a trademark registration, a trade dress drawing, or a mask work from a copyright registration.
11 . The device of claim 9 , wherein communicating the predicted infringement data to the recipient device comprises communicating the predicted infringement data to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement data.
12 . The device of claim 9 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice prompt to a chatbot.
13 . The device of claim 12 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice alert to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement notice alert.
14 . The device of claim 9 , wherein the computer-executable instructions, when executed by the processor, cause the processor to perform operations further comprising: identifying a domain of interest associated with the object, wherein the domain of interest comprises a type associated with the object; and determining, based on the domain of interest, an intellectual property classification associated with the type, wherein the intellectual property asset is identified from a plurality of intellectual property assets in the intellectual property classification.
15 . A non-invasive scanning device comprising: an image sensing device that is configured to capture images of a container; an emitter that is configured to emit energy toward the container; a processor; and a memory that stores computer-executable instructions that, when executed by the processor, cause the processor to perform operations comprising generating the images of the container using the image sensing device and the emitter, wherein the images of the container comprise a representation of an object; obtaining, from an intellectual property data source, an intellectual property asset image that is included in an intellectual property asset; performing, using a neural network, a coarse match operation to identify a coarse match comprising an image pair, the image pair comprising one image of the images of the container and the intellectual property asset image; applying an image modification operation to one of the intellectual property asset image or the one image of the images of the container to obtain an augmented intellectual property asset image or an augmented image of the container that is associated with the object; creating, using the neural network and based on the coarse match, a refined match comprising one of: a first pairing of the augmented intellectual property asset image and the one image of the images of the container, a second pairing of the intellectual property asset image and the augmented image of the container, or a third pairing of the augmented intellectual property asset image and the augmented image of the container; and in response to creating the refined match, communicating predicted infringement data to a recipient device via an application programming interface.
16 . The non-invasive scanning device of claim 15 , wherein the intellectual property asset image comprises a patent drawing from a patent, a photograph from a patent, an image from a trademark registration, a trade dress drawing, or a mask work from a copyright registration.
17 . The non-invasive scanning device of claim 15 , wherein communicating the predicted infringement data to the recipient device comprises communicating the predicted infringement data to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement data.
18 . The non-invasive scanning device of claim 15 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice prompt to a chatbot.
19 . The non-invasive scanning device of claim 15 , wherein communicating the predicted infringement data to the recipient device comprises communicating a predicted infringement notice alert to a recipient device controller, wherein the recipient device is configured to actuate a component of the recipient device in response to receiving the predicted infringement notice alert.
20 . The non-invasive scanning device of claim 15 , wherein the neural network is trained using an independent loss function to learn intellectual property classification information, and wherein the neural network is trained using a shared loss function to learn image similarity, and wherein the shared loss function uses contrastive or triplet loss for shared loss.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application is a continuation of and claims priority to U.S. patent application Ser. No. 16/707,702, entitled “Infringement Detection System,” filed Dec. 9, 2019, now U.S. Pat. No. 11,861,528, which is incorporated herein by reference in its entirety and which is a non-provisional of and claims priority to U.S. Prov. Pat. App. No. 62/777,088, entitled “Potential Infringement Detection,” filed Dec. 8, 2018, which is incorporated herein by reference in its entirety. BACKGROUND With the rapid global growth of ecommerce, it may be difficult for online marketplaces, sellers, buyers, customs officers, logistics professionals, and/or other entities to identify when a transaction, border crossing, or shipping activity involves an illegal counterfeit good, a knockoff product, and/or an infringing item for sale. Online sales growth may be driven by spontaneous purchases. Ecommerce driven by images (containing one or more trademarks, product designs, or graphical copyrighted content items) can occur when the image is first seen physically by a consumer and then acquired by a mobile device or other computing device for further search (e.g., capturing a picture on a camera or an augmented reality device and then using the image to search online for a product for sale); shared in an online store on a web page and offered for sale; displayed in a frame of video/animation/game with a call to action to purchase online; viewed through an image scan (e.g., x-ray, CT scan or other physical or medical imaging) that can trigger another computerized action or online transaction; or accessed in bulk by online store-generation tools to establish multiple online marketplace/website offerings. Determining that an image contains a fake or counterfeit (e.g., a product that infringes an intellectual property (“IP”) asset such as a trademark, a trade dress, a patent, a copyright, a mask work, or the like) may be difficult. In addition to accidental infringement of IP assets, some products may be knockoffs or pirated products (e.g., deliberate copies of IP assets such as design patents, trade dress registrations, trademarks, mask works, copyrighted works, or the like). Some attempts have been made to match trademark images to other images to detect infringement, but these attempts have not been considered very precise and/or accurate. Namely, these technologies have varying degrees of reliability (often measured as mean average precision or maP) and often do not consistently recognize trademarks in skewed or sharp perspectives. These technologies can be configured to match pixel-based tonal images to other tonal images, which may be considered “richer” than merely black-and-white line drawings and/or bitmapped images. These technologies, however, can fail and/or can produce poor predictive results when the input image is from a non-tonal domain, e.g., a line drawing, a sketch, or an engineering drawing image. Other technologies may attempt to match entire images and may not be properly trained to match one image within a more complex image. Furthermore, existing technologies can be defeated in a number of ways. In particular, a trademark may be hidden or removed from a potentially infringing product image to prevent matching from occurring. Similarly, images may be rotated, flipped, skewed, inverted or populated with visual noise to make matching difficult. A mark or feature in the image may appear similar to the human eye while not exactly matching an input mark that needs detection. Similarly, some trademark colors may be changed, creating an open question as to whether a match is viable. Determining that a potentially infringed image is, in fact, offered by an actually infringing, illicit, or otherwise unauthorized third party is a conclusion that is typically not reached by image matching systems. Currently, image matches are escalated to human analysts for a laborious manual evaluation that can include determining a) whether an image match is sufficient to generate concern and b) whether the party displaying, making, using, offering to sell, or selling the product in the image (or the image itself) has the rights to do so. This determination may often be made only from the viewpoint of the IP creator or owner and it therefore may be rarely if ever possible to make bulk IP infringer party determinations on behalf of an external party. SUMMARY The present disclosure is directed to an infringement detection system. The infringement detection system can be directed to multiple domains of interest, including images on the internet, in video, in animations, in live scanner feeds, in video streams, in games, in image scans (e.g., x-rays or CT scans), in images formed from LIDAR, in images captured on cameras, in images captured via augmented reality/mixed reality/virtual reality, or other image types. The concepts and technologies disclosed herein can utilize a database of potential infringing im