Patent Grant
10127717
This disclosure relates generally to the field of representing a graphical object display modeling using mathematical algorithms. More particularly, the disclosure relates to creating three-dimensional models of clothing and other wearables.
Description
One of the basic needs of human beings is clothing. In 2016, annual apparel sales were expected to exceed $1.4 trillion globally, and more than $300 billion in the United States. In 2015, apparel became the highest selling category in online retail within the United States, reaching about $75 billion in annual sales.
There are advantages in selling and purchasing clothing online. From the customers point of view, the time and effort to travel to the store may be avoided. They may purchase the clothing directly on their computer wherever they happen to be. From the retailer's point of view, the need for brick and mortar stores may be avoided altogether or reduced in number, thereby potentially saving retail store rents. The self-service nature of online sales may reduce the need for retail sales staff, and the associated costs.
There may also be disadvantages in selling and purchasing clothing online. In contrast to a customer who visits a brick-and-mortar retail store, an online customer cannot physically try on clothing before purchase. The lack of pre-sales fitting may increase the return rate because customers find that they are not satisfied with the fit or look of the clothing when they receive the order. The extra shipping and handling cost and effort of returned items can discourage purchases in the online sales channel, increase wear on unsold merchandise, and reduce profits.
E-commerce websites often display images of clothing dressed on mannequins or human models so that potential customers can visualize the clothes as they would be worn. These images are often generated by the manufacturer or the retailer to promote the product. These photographed professional models that may not represent the body type of the typical customer. However, many retailers also take photos of so-called plus-sized models and others that may represent more typical customers. These efforts may help potential customers better visualize the clothing on themselves.
When a potential customer visits a retail store, they typically have the option to try on in-stock clothing. They can verify which size of any such garments fit them and can see how the selected garment looks on them. They can try on clothing and get immediate feedback from the friends or family who visit the store with them. However, online purchases are fulfilled through centralized warehouses that may provide an advantage in terms of consolidated inventory. Consolidated inventory may allow a wider range of stocking units (SKUs) and sizes than can cost-effectively be maintained in inventory in each of numerous smaller retail stores.
Clothing retailers often provide generous return policies for clothing that is purchased online. This may encourage consumers to make online purchases of clothing despite the lack of a pre-sales fitting process.
A system and method for capturing a 3D clothing model is disclosed.
The system captures a three-dimensional point cloud of the reference mannequin and a three-dimensional point cloud of the reference mannequin wearing an article of clothing or other wearable.
In some embodiments, the scan information is captured using at least one depth sensor. The depth sensors measure distance from the sensor to the surface of the naked or clothed reference mannequin. The reference mannequin or the sensor or both are moved in a controlled manner so that the sensor can capture surface positions of the entire surface of the naked or clothed reference mannequin. A processor processes the scan information using well-known techniques for depth sensor information to produce a point cloud representing the surface of the naked or clothed mannequin.
In some embodiments, the scan information is captured using at least one image sensor. The image sensor captures 2D images of the surface of the naked or clothed reference mannequin. The reference mannequin or the sensor or both are moved in a controlled manner so that the sensor can capture 2D images of the naked or clothed reference mannequin from various perspectives. A processor processes the scan information using well-known techniques for 2D images to produce a point cloud representing the surface of the naked or clothed mannequin.
The point cloud for the wearable and the point cloud for the reference mannequin are align using at least one key feature common to both point clouds. Key features may include at least one of shoulders, neck, armpits, and crotch.
The point clouds are sliced perpendicular to lengthwise axes and centroids are computed for each slice. In a preferred embodiment, the point clouds are represented in polar co-coordinates for processing.
Difference tables are generated by computing the distances between corresponding points in the point cloud for the wearable and the point cloud for the reference mannequin. Each entry in the difference table corresponds to a distance that is associated with a specific point in the point cloud of the reference mannequin. Color and texture information may also be associated which each difference. This color and texture represents the color and texture of the wearable.
In a preferred embodiment, these differences are computed in parallel for each slice using one or more graphics processing units.
Three-dimensional display modeling has been used to render images of people wearing clothing. Technicians use high-end 3-D animation software to manually create a library of clothing articles. Physical simulation of the clothing design and materials are then applied to the model to render an image of the clothing article worn by a person.
These methodologies can require expensive highly skilled technicians to run the animation software. The work can be labor-intensive and time-consuming. The physical simulation and rendering can require substantial computer resources. In some cases, these methodologies can take 20-60 minutes on the fastest cloud processor.
Given the number of stock keeping units (SKUs) in a clothing line, including variations in sizes and materials, the significant skilled labor, time, and compute resources to generate 3D models for each clothing line would be substantial.
The simple scan process described herein does not require operators skilled in computer graphics or physical modeling. There does not need to be any complex physical simulation as to how the wearable will look on the model of the human body. The position of the surface of the wearable on the reference model is captured as it is actually positioned on the reference mannequin during the scan operation. An operator might just press a button to scan a reference mannequin, dress the reference mannequin with a wearable and press the button again to scan the dressed mannequin. A difference table for that wearable could be automatically created.
Prior art systems use complex graphical representations of both the mannequin and the wearable that necessitates significant complex computations to render as an image of the mannequin wearing the wearable. In some embodiments, this may take on the order of 20-60 minutes per rendering.
In contrast, the difference tables are simply tables of scalar adjustments to associated points in a point cloud for the reference mannequin. Because of the simplified computations required, and the facilitation of parallel processing of each slice, each clothing article might take on the order of 3-5 minutes.
In one embodiment, the difference tables for wearables are used with as disclosed in U.S. non-provisional application entitled “System for Virtually Trying On Clothing”, filed on Feb. 17, 2017, Ser. No. 15/436,514.
Various examples of embodiments will be described below with reference to the drawings. The following exemplary embodiments are illustrative and are not to be construed as limiting.
A sensing device 104 is configured to receive information through a sensing input 105 under the control of a processor 107.
In some embodiments, the sensing device 105 is a depth sensor configured to measure the distance from the sensing input 105 to the surface of the reference mannequin 102 at one or more surface points within the field of view 106 of the sensing input 105.
In some embodiments, the sensing device 104 includes a sensing input 105 that has a single depth sensor that is moved in a controlled manner around the periphery of the mannequin 102 to capture the depth information from multiple perspectives around the mannequin 102. By tracking the relative motion between the sensing input 105 and the mannequin 102 for each of the depth measurements, the sensing device 104 can determine surface position information at various points on the reference mannequin 102. This information can be processed to produce a point cloud representing the scanned portion of the reference mannequin 102. Collectively, point cloud information represents the surface shape of the mannequin 102.
In other embodiments, the sensing device 104 remains in a fixed position and the base 101 is rotated through a 360° rotation in synchronization with the sensing input 105 of the sensing device 104 to gather the depth information from multiple perspectives around the mannequin 102.
In yet other embodiments, both the base 101 and the sensing device 104 are configured to move to gather the depth information from multiple perspectives around the mannequin 102.
In some embodiments, the sensing device 104 surrounds the mannequin 102 and the sensing input 105 includes an array of depth sensors dispersed around the mannequin 102 to capture all the depth information from multiple perspectives around the mannequin 102 in parallel.
Between the extremes of capturing all depth sensor information in parallel, and capturing each depth sense measurement in sequence, there may be smaller arrays of depth sensors that are moved around the reference mannequin 102 to capture the depth sensor information. More depth sensors allow more scanning to be performed in parallel which may lead to faster scan results. However, more depth sensors may require larger and more expensive scanning devices.
A scan information 111 is received by a processor 107. The scan information 111 includes each of the multiple depth measurements and the position of the associated sensor for each of the measurements. The scan information 111 may also include color and texture information at each of the points. The processor 107 processes the first scan information 111 according to well-known methods for processing depth information to generate a first point cloud 108 and stores the first point cloud 108 in a database 110. The first point cloud 108 is a constellation of points indicating the relative position of the surface of the mannequin 102 and color and/or texture information at those points as derived from the scan information 111.
In an alternative embodiment, the sensing device 104 is a two-dimensional (2D) image sensor that captures an image of the mannequin 102 from various perspectives around the mannequin 102 as described with reference to the depth sensor(s).
The first scan information 111 is received by the processor 107. The first scan information 120 includes each of the multiple 2D images and the associated perspectives of the one or more image sensors. The processor 107 processes the scan information 111 according to well-known methods for processing 2D image information to generate a point cloud 108 and transmits the first point cloud 108 into a database 110. The point cloud 108 is a constellation of points indicating the relative position of the surface of the mannequin 102 and color and/or texture information at those points as derived from the scan information 120.
It will be apparent that other schemes for moving one or more sensors around the periphery of the mannequin 102 may be used to generate the scan information 111.
The sensing device 104 is operated as described with reference to
In embodiments of the system 100 that use one or more depth sensors, the second scan information 112 includes each of the multiple depth measurements and the associated perspectives for each of the measurements. The processor 107 receives the scan information 112 and processes the scan information 112 according to well-known methods for processing depth information to generate a point cloud 109 and transmits the first point cloud 107 over a bus into a database 109. The point cloud 109 is a constellation of points indicating the relative position of the surface of the mannequin 102 wearing the dress 103 and color and/or texture information at those points as derived from the scan information 112.
In embodiments of the system 100 using one or more 2D image sensors, the scan information 112 includes each of the multiple images and the associated perspectives of each of the Images. The processor 106 processes the second scan information 120 according to well-known methods for processing 2D image information to generate a second point cloud 108 and transmits the second point cloud 108 into a database 109. The second point cloud 108 is a constellation of points indicating the relative position of the surface of the mannequin 102 and color and/or texture information at those points as derived from the scan information 112.
In step 300, the reference mannequin 102 is scanned to generate the scan information 111 as described with reference to
In step 305, the processor 107 processes the scan information to generate the point cloud 108 as described with reference to
In Step 310, the reference mannequin 102 is dressed with a wearable.
In some embodiments, the wearable is one of various types of clothing such as pants, shorts, dresses, skirts, shirts, blouses, stockings, gloves, hats, and the like. In some embodiments, the wearables include accessories such as eyeglasses, sunglasses, rings necklaces and earrings.
Generally, the process is performed with a single wearable for each scan so that the model generated by this process is associated with the individual wearable. In some cases, however, more than one wearable may be processed together as part of a set, such as a two-piece bikini.
In step 315, the reference mannequin 102 is scanned to generate the scan information 112 as described with reference to
In step 320, the processor 107 receives the scan information 112 and processes the scan information 112 to generate the point cloud 109 as described with reference to
In step 325, the point cloud 109 is aligned with the point cloud 108.
The point cloud 109 is a 3D surface representation of the reference mannequin 102 wearing the wearable. The point cloud 108 is a 3D surface representation of the reference mannequin 102 without the wearable.
In some embodiments, the point cloud 109 is aligned with the point cloud 108 by aligning key features in both point clouds. Key features may include the shoulder, neck, armpit or crotch, for example. Pattern recognition may be used to locate these key features in each of the point cloud 109 and the point cloud 108 even on a point cloud which covers some of these key features.
In step 330 the aligned point clouds are separated into slices by the various feature of the reference mannequin. Separating the point cloud into slices allows for more effective processing, including parallel processing by graphics processing units.
The body is generally sliced perpendicular to the lengthwise axis for each body feature. For example, the slices in the arms are sliced perpendicular to the lengthwise axis of each arm. The slices in the legs are cut perpendicular to the axis along the lengthwise axis of each leg. The slices in the torso are cut perpendicular to the lengthwise axis of each trunk. The slices in the torso are cut perpendicular to the lengthwise axis of each trunk. Similarly, the head, hands and feet are cut along the associated lengthwise axis.
The sizes of the slices may vary depending on the relative size of the feature and detail required for that feature of the body. For example, the fingers may be split into smaller slices that the legs.
Centroid are computed for each slice. In a preferred embodiment, the point clouds are represented in polar coordinates. However, other representation systems may be used.
In step 345, a smoothing filter is applied to the centroids. The filter removes discontinuities and other irregularities in the positing of the slices along the lengthwise axis of each portion of the point cloud.
In step 340, a table of entries having distances between corresponding points in the first and second point cloud are generated and stored in a database.
In step 345, it is determined whether another wearable will be scanned during this batch process. If so, the currently worn wearable is removed and the process continues at step 310 using the same reference point cloud that was initially captured in steps 300-305.
In some embodiments, similar processing is perform used only a reference body representing only a portion of a human body. For example, a reference body limited to the portion of the body below the waist might be used to create 3D model of pants and skirts. A reference body limited to the torso might be used to create 3D models of shirts and blouses. A reference body limited to one or both legs might be used to create 3D models of stockings or socks.
In some embodiments, the point clouds include polygon meshes with color and texturing.
A lengthwise axis 200 passes through the centroids at the center of each slice.
A point 201a is a point at the boundary between slices. Only a few sample points are shown in the figure. In some embodiments, a slice may have approximately five thousand points dispersed around the sidewall of the slice to represent the surface of the scanned object.
A point 201b represents the same point on the scanned object as the point 201a but it is assigned to the slice 220. The point 201a and the point 201b are associated with each other by a connection link 201c.
Similarly, a point 202a and a point 202b represent the same point on the scanned object but the point 202a is assigned to the slice 210 and the point 202b assigned to the slice 220. The point 202a and the point 202b are associated with each other by a connection link 202c.
Similarly, a point 205a and a point 205b represent the same point on the scanned object but the point 205a is assigned to the slice 210 and the point 202b assigned to the slice 220. The points 205a and 205b are associated with each other by a connection link 205c.
Similar relationships are created at the boundary between the slice 220 and the slice 230. A point 211b represents the same point on the scanned object as a point 211a. Point 211a is assigned to the slice 220 and the point 211b is assigned to the slice 230. The point 211a and the point 211b are associated with each other by a connection link 211c.
Similarly, a point 215a and a point 215b represent the same point on the scanned object but the point 215a is assigned to the slice 220 and the point 215b is assigned to the slice 230. The point 215a and the point 215b are associated with each other by a connection link 215c.
A scanned object may be represented by long sequences of slices connected using the illustrated principals.
Collectively, the connection links make up a connection map that capture the relationships between the points at the edges between the slices so that the slices can be reassembled after processing. The slices are reassembled by bringing the slices together and removing each redundant point linked at the boundary.
In some embodiments, the point cloud shown represents a portion of the reference mannequin. In other embodiments, the point cloud shown represents a portion of the wearable being worn by the reference mannequin.
In a preferred embodiment, the point cloud of the wearable and the point cloud for the reference mannequin are aligned and sliced as described herein so that each slice for the wearable has a corresponding slice for the reference mannequin. The corresponding slices are associated with the same portion of the reference mannequin. These corresponding slices can be processed together independently of the other slices thereby enabling parallel processing for better performance.
The segment of the point cloud 580 may represent a portion of a leg of the reference mannequin and the segment of the point cloud 590 may represent the leg portion of a pair of pants.
Only a cross-section portion of the point cloud 590 is shown. It is common for the point cloud of the wearable to wrap around the point cloud 580. The point cloud 590 is also separated into slices. The portion of the point cloud 590 that wraps in front of and behind the segment of the point cloud 580 is not shown to clearly illustrate the computation of a difference table representing the 3D positions of the point cloud 590 as offset from the point cloud 580.
The slices (not shown) for the point cloud 590 are aligned with the slices of the point cloud 580 using key features common to both point clouds as described herein.
These point clouds are shown with the slices assembled because that is the positioning which is analogous to the real-world scenario of a wearable wrapped around a portion of a reference mannequin. Thus, a human is better able to intuitively understand the relationships discussed during the calculations. However, the following calculations are generally performed on slices that have been separated as shown in
A point 501b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 501a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 510. The distance between the point 501b and the point 501a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
Similarly, on the opposite side of the point cloud 580, a point 505b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 505a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 510. The distance between the point 501b and the point 501a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
A point 511b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 511a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 550. The distance between the point 511b and the point 511a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
Similarly, on the opposite side of the point cloud 580, a point 515b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 515a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 550. The distance between the point 515b and the point 515a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
There are numerous other points in the point cloud 580, not only at the boundaries between the slices, but along the sides within each slice to define the contours of the surface along the side walls of each slice. Similarly, there are numerous other points in the point cloud 590, not only at the boundaries between the slices, but along the sides within each slice to define the contours of the surface along the side walls of each slice. A scalar distance is computed between the corresponding points in the point cloud 590 and the point cloud 580 and stored in a table and each distance is associated with each slice and point of the reference mannequin. These distances are also referred to as a difference, and the table is sometimes referred to as a difference table.
The segment of the point cloud 580 may represent a portion of the reference mannequin and the segment of the point cloud 590 may represent a portion of a wearable 591.
Only a cross-section portion of the point cloud 591 is shown. It is common for the point cloud of the wearable to wrap around the point cloud 580. The point cloud 591 is also separated into slices. The portion of the point cloud 591 that wraps in front of and behind the segment of the point cloud 580 is not shown to clearly illustrate the computation of a difference table representing the 3D positions of the point cloud 591 as offset from the point cloud 580.
The slices (not shown) for the point cloud 591 are aligned with the slices of the point cloud 580 using key features common to both point clouds as described herein.
These point clouds are shown with the slices assembled because that is the positioning which is analogous to the real-world scenario of a wearable wrapped around a portion of a reference mannequin. Thus, a human is better able to intuitively understand the relationships discussed during the calculations. However, the following calculations are generally performed on slices that have been separated as shown in
The point cloud 591 is for a simplified wearable not to scale with reference to the slices. It is primarily configured here to illustrate various exemplary relationships between the point cloud for a wearable and the point cloud for the reference mannequin for the purposes of computation of a difference table.
The cross-section portion of the point cloud 591 does not extend over the portion of the slice 510. The point cloud 580 has a point 501a and a point 505a as derived from the scan process of the reference mannequin. When the wearable is scanned, the surface points captured at the slice 510 will be the same as that captured for the naked reference mannequin since the mannequin is naked at this slice. In a preferred embodiment, the system determines whether points such as the point 501a is included within the point cloud for the 591 and the point cloud for 580, or such corresponding points are within a certain minimum scanner resolution so as to be treated as the same point. The difference between these corresponding points in the point cloud 591 and the point cloud 580 is zero. In a preferred embodiment, an entry in the difference table is not created for such points.
A point 501b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 501a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 510. The distance between the point 501b and the point 501a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
Similarly, on the opposite side of the point cloud 580, a point 505b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 505a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 510. The distance between the point 501b and the point 501a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
A point 511b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 511a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 550. The distance between the point 511b and the point 511a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
Similarly, on the opposite side of the point cloud 580, a point 515b in the point cloud 590 represents a point on the surface of the wearable 109 at a slice (not shown) that is aligned with a point 515a in the point cloud 590 that represents a point on the surface of the reference mannequin 102 at the slice 550. The distance between the point 515b and the point 515a represents the distance that the surface of the wearable 109 is positioned from the surface of the underlying reference mannequin.
There are numerous other points in the point cloud 580, not only at the boundaries between the slices, but along the sides within each slice to define the contours of the surface along the side walls of each slice. Similarly, there are numerous other points in the point cloud 590, not only at the boundaries between the slices, but along the sides within each slice to define the contours of the surface along the side walls of each slice. A scalar distance is computed between the corresponding points in the point cloud 590 and the point cloud 580 and stored in a table and each distance is associated with each slice and point of the reference mannequin. These distances are also referred to as a difference, and the table is sometimes referred to as a difference table.
In some embodiments, the machine 900 is the system 150 according to an embodiment as described herein or a component of such systems, such as one or more processors that make up the system 150. In other embodiments, the machine 900 is the database system 110 according to an embodiment as described herein.
The machine 900 includes a processor 960 (e.g., a central processing unit (CPU), a graphics processing unit (GPU) or both), a main memory 970 and a nonvolatile memory 980, which communicate with each other via a bus 902. In some embodiments, the machine 900 may be a cluster of computers or comprise multiple processors or multiple processor cores. In one embodiment, the machine 900 also includes a video display 910, an alphanumeric input device 920 (e.g., a keyboard), a cursor control device 930 (e.g., a mouse), a drive unit 940 (e.g., solid state drive (SSD), hard disk drive, Digital Versatile Disk (DVD) drive, or flash drive), a sensing device 950 (e.g., a speaker) and a network interface device 990.
In some embodiments, the machine 900 is includes the sensing device 104.
In some embodiments, the video display 910 includes a touch-sensitive screen for user input. In some embodiments, the touch-sensitive screen is used instead of a keyboard and mouse. The drive unit 940 includes a machine readable medium 942 on which is stored one or more sets of instructions 944 (e.g. software) embodying any one or more of the methods or functions of the inventive subject matter.
The instructions 944 may also reside, completely or partially, on machine-readable media within the main memory 940 and within machine-readable media within the processor 960 during execution thereof by the machine 900. The instructions 944 may also be transmitted or received over a network 995 via the network interface device 990. In some embodiments, the main memory 970 and the machine-readable medium 942 also includes a data 946 including the scan information or the point clouds.
While the machine-readable medium 942 is shown in an exemplary embodiment to be a single medium, the term “machine-readable medium” should be taken to include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) that store the one or more sets of instructions and/or data. The term “machine readable medium” shall also be taken to include any non-transitory medium that is capable of storing, encoding or carrying a set of instructions for execution by the machine and that cause the machine to perform any one or more of the methods or functions of the inventive subject matter. The term “machine-readable medium” shall accordingly be taken to include, but not be limited to, solid-state memories, optical and magnetic media, and other non-transitory tangible media.
In general, the methods executed to implement the embodiments of the disclosure, may be implemented as part of an operating system or a specific application, component, program, object, module or sequence of instructions referred to as “programs.” For example, one or more programs may be used to execute specific processes according to the inventive subject matter. The programs typically comprise one or more instructions set at various times in various memory and storage devices in the machine, and that, when read and executed by one or more processors, cause the machine to perform operations to execute methods, functions and other elements of the inventive subject matter.
Moreover, while embodiments have been described in the context of machines, those skilled in the art will appreciate that the various embodiments are capable of being distributed as a program product in a variety of forms, and that the disclosure applies equally regardless of the particular type of machine or computer-readable media used to actually effect the distribution. Examples of machine-readable media include, but are not limited to, recordable type media such as volatile and non-volatile memory devices, solid state drives (SSDs), flash memory devices, floppy and other removable disks, hard disk drives, and optical disks such as Compact Disk Read-Only Memory (CD-ROMS) and Digital Versatile Disks (DVDs), among others.
The torso portion of the point cloud for an avatar might look very similar except that the color and texture of the surface of the avatar would be different than that of the mannequin.
The torso portion of the point cloud for an avatar dressed with a blouse might look very similar except that the color and texture of the surface of the avatar would be different than that of the mannequin.
This U.S. non-provisional application claims the benefit of U.S. provisional application No. 62/296,001, filed on Feb. 16, 2016, the contents of which are expressly incorporated by reference herein. This U.S. non-provisional application is related to the following commonly-owned U.S. patent applications, which are hereby expressly incorporated by reference in their respective entireties: (1) U.S. non-provisional application entitled “System and Method for Virtually Trying-On Clothing”, filed on Feb. 17, 2017, Ser. No. 15/436,514, which claims the benefit of U.S. provisional application No. 62/296,005, filed on Feb. 16, 2016;(2) U.S. non-provisional application entitled “Virtually Sharing Customized Clothing”, filed on Feb. 17, 2017, Ser. No. 15/436,580, which claims the benefit of U.S. provisional application No. 62/296,008, filed on Feb. 16, 2016.(3) U.S. non-provisional application entitled “System and Method for Targeted Personalized Ads”, filed on Feb. 17, 2017, Ser. No. 15/436,547, which claims the benefit of U.S. provisional application No. 62/296,013, filed on Feb. 16, 2016.
| Number | Name | Date | Kind |
|---|---|---|---|
| 5680528 | Korszun | Oct 1997 | A |
| 5822466 | Morfill et al. | Oct 1998 | A |
| 5850222 | Cone | Dec 1998 | A |
| 5923780 | Morfill et al. | Jul 1999 | A |
| 5930769 | Rose | Jul 1999 | A |
| 6009212 | Miller et al. | Dec 1999 | A |
| 6182871 | Lam | Feb 2001 | B1 |
| 6307568 | Rom | Oct 2001 | B1 |
| 6310627 | Sakaguchi | Oct 2001 | B1 |
| 6404426 | Weaver | Jun 2002 | B1 |
| 6415199 | Liebermann | Jul 2002 | B1 |
| 6546309 | Gazzuolo | Apr 2003 | B1 |
| 6556196 | Blanz et al. | Apr 2003 | B1 |
| 6611630 | Miller et al. | Aug 2003 | B1 |
| 6888965 | Rath et al. | May 2005 | B1 |
| 6901379 | Balter et al. | May 2005 | B1 |
| 7133839 | Inoue et al. | Nov 2006 | B2 |
| 7149665 | Feld et al. | Dec 2006 | B2 |
| 7328177 | Lin-Hendel | Feb 2008 | B1 |
| 7355597 | Laidlaw et al. | Apr 2008 | B2 |
| 7398133 | Wannier et al. | Jul 2008 | B2 |
| 7433753 | Okada et al. | Oct 2008 | B2 |
| 7492934 | Mundy et al. | Feb 2009 | B2 |
| 7978887 | Mundy et al. | Jul 2011 | B2 |
| 8170919 | Barbour et al. | May 2012 | B2 |
| 8275590 | Szymczyk et al. | Sep 2012 | B2 |
| 8290568 | Pichler et al. | Oct 2012 | B2 |
| 8359247 | Vock | Jan 2013 | B2 |
| 8364561 | Wolper et al. | Jan 2013 | B2 |
| 8427656 | Hullin et al. | Apr 2013 | B2 |
| 8438081 | Gray et al. | May 2013 | B2 |
| 8515145 | Mundy et al. | Aug 2013 | B2 |
| 8605148 | Robertson | Dec 2013 | B2 |
| 8700477 | Wolper et al. | Apr 2014 | B2 |
| 8711175 | Aarabi | Apr 2014 | B2 |
| 8818883 | Lawrence et al. | Aug 2014 | B2 |
| 8942456 | Hesthaven et al. | Jan 2015 | B2 |
| 9147207 | Haaramo et al. | Sep 2015 | B2 |
| 9189886 | Black et al. | Nov 2015 | B2 |
| 9191579 | Seidel et al. | Nov 2015 | B2 |
| 9241184 | Weerasinghe | Jan 2016 | B2 |
| 9292967 | Black et al. | Mar 2016 | B2 |
| 9489744 | Black et al. | Nov 2016 | B2 |
| 9576198 | Mundy et al. | Feb 2017 | B2 |
| 9679409 | Black et al. | Jun 2017 | B2 |
| 9710964 | Loper et al. | Jul 2017 | B2 |
| 9761060 | Black et al. | Sep 2017 | B2 |
| 9799098 | Seung et al. | Oct 2017 | B2 |
| 9898848 | Black et al. | Feb 2018 | B2 |
| 9961328 | Wu et al. | May 2018 | B2 |
| 10002460 | Black et al. | Jun 2018 | B2 |
| 10008007 | Taubin et al. | Jun 2018 | B2 |
| 10032254 | Harmeling et al. | Jul 2018 | B2 |
| 20010026272 | Feld et al. | Oct 2001 | A1 |
| 20010042029 | Galvez | Nov 2001 | A1 |
| 20030065578 | Peyrelevade et al. | Apr 2003 | A1 |
| 20030076318 | Shaw-Weeks | Apr 2003 | A1 |
| 20030101105 | Vock | May 2003 | A1 |
| 20040049309 | Gardner et al. | Mar 2004 | A1 |
| 20040083142 | Kozzinn | Apr 2004 | A1 |
| 20040227752 | McCartha et al. | Nov 2004 | A1 |
| 20050131776 | Perotti et al. | Jun 2005 | A1 |
| 20050267614 | Looney et al. | Dec 2005 | A1 |
| 20060202986 | Okada et al. | Sep 2006 | A1 |
| 20060287877 | Wannier et al. | Dec 2006 | A1 |
| 20070005174 | Thomas | Jan 2007 | A1 |
| 20070057942 | Unal | Mar 2007 | A1 |
| 20070130020 | Paolini | Jun 2007 | A1 |
| 20070179867 | Glazer et al. | Aug 2007 | A1 |
| 20070198120 | Wannier et al. | Aug 2007 | A1 |
| 20080249897 | Oh et al. | Oct 2008 | A1 |
| 20090018926 | Buehlman | Jan 2009 | A1 |
| 20090089186 | Paolini | Apr 2009 | A1 |
| 20090115777 | Reyers Moreno | May 2009 | A1 |
| 20100094729 | Gray et al. | Apr 2010 | A1 |
| 20100191770 | Cho et al. | Jul 2010 | A1 |
| 20100238271 | Pfeiffer | Sep 2010 | A1 |
| 20100306082 | Wolper et al. | Dec 2010 | A1 |
| 20110022965 | Lawrence et al. | Jan 2011 | A1 |
| 20110078055 | Faribault et al. | Mar 2011 | A1 |
| 20110298897 | Sareen et al. | Dec 2011 | A1 |
| 20120022978 | Manea et al. | Jan 2012 | A1 |
| 20120136755 | Yang | May 2012 | A1 |
| 20120299912 | Kapur et al. | Nov 2012 | A1 |
| 20130018763 | Ajala | Jan 2013 | A1 |
| 20130083065 | Schulze | Apr 2013 | A1 |
| 20130173226 | Reed et al. | Jul 2013 | A1 |
| 20130215116 | Siddique et al. | Aug 2013 | A1 |
| 20130249908 | Black et al. | Sep 2013 | A1 |
| 20130293686 | Blow et al. | Nov 2013 | A1 |
| 20140006386 | Oh | Jan 2014 | A1 |
| 20140022238 | Oh | Jan 2014 | A1 |
| 20140089135 | Linh et al. | Mar 2014 | A1 |
| 20140108208 | Piana | Apr 2014 | A1 |
| 20140149264 | Satyanarayana et al. | May 2014 | A1 |
| 20140176565 | Adeyoola et al. | Jun 2014 | A1 |
| 20140225888 | Bell et al. | Aug 2014 | A1 |
| 20140368499 | Kaur | Dec 2014 | A1 |
| 20150213646 | Ma | Jul 2015 | A1 |
| 20160035142 | Nair et al. | Feb 2016 | A1 |
| 20160086353 | Lukac | Mar 2016 | A1 |
| 20160148436 | Chang | May 2016 | A1 |
| 20160150217 | Popov | May 2016 | A1 |
| 20170214943 | Cohen | Jul 2017 | A1 |
| 20170249783 | Mach Shepherd | Aug 2017 | A1 |
| Number | Date | Country |
|---|---|---|
| 2462564 | Apr 2014 | EP |
| Entry |
|---|
| Protopsaltou, D., Luible, C., Arevalo, M., & Magnenat-Thalmann, N. (2002). A body and Garment Creation Method for an Internet Based Virtual Fitting Room. In Advances in Modelling, Animation and Rendering (pp. 105-122). Springer London. https://doi.org/10.1007/978-1-4471-0103-1_7. |
| Apparel Highest-Selling Category in E-Commerce in 2015. (Apr. 4, 2016). Retrieved Feb. 11, 2017, from https://sourcingjournalonline.com/apparel-was-highest-selling-category-in-online-retail-in-2015/. |
| Cordier, Frédéric, Wonsook Lee, Hyewon Seo, and Nadia Magnenat-Thalmann. “Virtual-try-on on the web.” Laval Virtual (2001). |
| Weiss, A., Hirshberg, D., & Black, M. J. (2011). Home 3D body scans from noisy image and range data. In 2011 International Conference on Computer Vision. IEEE. https://doi.org/10.1109/iccv.2011.6126465. |
| Number | Date | Country | |
|---|---|---|---|
| 20170287210 A1 | Oct 2017 | US |
| Number | Date | Country | |
|---|---|---|---|
| 62296001 | Feb 2016 | US |
