DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of the Claims
Claims 1-20, as originally filed, are currently pending and have been considered below.
Drawings
Color photographs and color drawings are not accepted in utility applications unless a petition filed under 37 CFR 1.84(a)(2) is granted. Any such petition must be accompanied by the appropriate fee set forth in 37 CFR 1.17(h), one set of color drawings or color photographs, as appropriate, if submitted via the USPTO patent electronic filing system or three sets of color drawings or color photographs, as appropriate, if not submitted via the via USPTO patent electronic filing system, and, unless already present, an amendment to include the following language as the first paragraph of the brief description of the drawings section of the specification:
The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.
Color photographs will be accepted if the conditions for accepting color drawings and black and white photographs have been satisfied. See 37 CFR 1.84(b)(2).
Note that the requirement for three sets of color drawings under 37 C.F.R. § 1.84(a)(2)(ii) is not applicable to color drawings submitted via the USPTO patent electronic filing system. Therefore, only one set of such color drawings is necessary when filing via the USPTO patent electronic filing system.
The drawings are objected to because Figures 1-5 and 7 include color, but no petition under 37 C.F.R. § 1.84(a)(2) has been filed.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1, 4, 5, 10, 11, 13, 16, 19 and 20 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Wang W, Chen Y, Liu Y, Yuan Q, Yang S, Zhang Y. MVOC: a training-free multiple video object composition method with diffusion models. arXiv preprint arXiv:2406.15829. 2024 Jun 22, hereinafter, “Wang”.
As per claim 1, Wang discloses a method comprising:
generating, by a processing device, mask data that separates a subject depicted in frames of a subject video from a first environment and foreground feature data describing features of the subject (Wang, page 4, Figure 2. Multiple video object composition framework. Our method presents a two-stage approach: video object preprocessing and generative video editing. In the preprocessing stage, we perform … object extraction … as well as mask extraction; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, The proposed multiple video object composition framework is illustrated in Fig. 2 ... the video objects are derived to the corresponding object masks {Myi}N, i=1 by the mask extraction module ... we combine the video objects and the corresponding masks into a new conditional guidance video);
receiving, by the processing device, a condition frame depicting a second environment, the second environment being different than the first environment (Wang, page 4, Figure 2. Multiple video object composition framework. we edit the first frame by an image editing model, then use video object dependence for conditional guidance video generation);
generating, using a machine-learning model with inputs of the foreground feature data and the mask data, a composite video that aligns movement of the subject with the second environment, the machine-learning model using the condition frame to generate and condition a depiction of the second environment in the composite video (Wang, page 1, Abstract, video object composition tasks ... ensure that the objects in the composited video maintain motion and identity consistency ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, The proposed multiple video object composition framework is illustrated in Fig. 2 ... the video objects are derived to the corresponding object masks {Myi}N, i=1 by the mask extraction module ... we combine the video objects and the corresponding masks into a new conditional guidance video … we leverage an image editing model to obtain an edited first frame … and feed it into the I2V generation model for video generation … we use the features and attention maps of the corresponding noise {{yit}N, i =1}T, t=1 as condition injections to generate the composited video); and
presenting, by the processing device, the composite video via a user interface (Wang, pages 7-8, 5.3. Quantitative results, User study. 25 participants are invited to vote on seven videos composited ... we showed the participants three input videos and two composited videos generated by our method and a compared method, and then recorded the number of participants voting on the harmony and temporal consistency of the composited videos).
As per claim 4, Wang discloses the method of claim 1, wherein the method further comprises generating, using an image encoder, a feature representation of the condition frame with background feature data being a last hidden layer of the feature representation, the background feature data being an input to the machine-learning model (Wang, page 3, 3.3. Attention mechanisms, The denoising model ϵθ (xt, t) is often instantiated by architectures like UNet, which has four basic modules: convolutional modules, spatial self-attention modules, spatial cross-attention modules, and temporal self-attention modules. In this work, we focus on latent diffusion models [33] for video generation, which formulates the attention operation as: [5a; 5b], where z is the input hidden state to the attention layer and WQ, WK and WV are trainable projection matrices that map z onto query, key and value vectors, respectively; Wang, page 4, Figure 2. Multiple video object composition framework; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, pages 3-4, 4. The proposed method, 4.1. Overall framework, autoencoder that compresses a RGB pixel space to a low-resolution latent space and reconstructs the latent space back to RGB frames in latent diffusion models).
As per claim 5, Wang discloses the method of claim 4, wherein: the machine-learning model is a convolutional neural network; and the background feature data are injected through cross-attention layers of a denoising U-Net of the convolutional neural network (Wang, page 3, 3.3. Attention mechanisms, The denoising model ϵθ (xt, t) is often instantiated by architectures like UNet, which has four basic modules: convolutional modules, spatial self-attention modules, spatial cross-attention modules, and temporal self-attention modules. In this work, we focus on latent diffusion models [33] for video generation, which formulates the attention operation as: [5a; 5b], where z is the input hidden state to the attention layer and WQ, WK and WV are trainable projection matrices that map z onto query, key and value vectors, respectively; Wang, page 4, Figure 2. Multiple video object composition framework).
As per claim 10, Wang discloses the method of claim 1, wherein the condition frame includes a digital photograph of the second environment, a frame of a video depicting the second environment, or a digital image of the second environment generated using another machine-learning model or photo editing resources (Wang, page 4, Figure 2. Multiple video object composition framework. we edit the first frame by an image editing model, then use video object dependence for conditional guidance video generation).
As per claim 11, Wang discloses a computing device comprising:
a processing device; and a computer-readable storage medium storing instructions that, responsive to execution by the processing device (Wang, pages 5-6, 5. Experiments, 5.1. Experimental setup, All experiments are performed on a single NVIDIA A6000 GPU), causes the processing device to perform operations including:
generating mask data that separates a subject depicted in frames of a subject video from a first environment and foreground feature data describing features of the subject (Wang, page 4, Figure 2. Multiple video object composition framework. Our method presents a two-stage approach: video object preprocessing and generative video editing. In the preprocessing stage, we perform … object extraction … as well as mask extraction; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, The proposed multiple video object composition framework is illustrated in Fig. 2 ... the video objects are derived to the corresponding object masks {Myi}N, i=1 by the mask extraction module ... we combine the video objects and the corresponding masks into a new conditional guidance video);
generating background feature data from a condition frame depicting a second environment, the second environment being different than the first environment (Wang, page 4, Figure 2. Multiple video object composition framework. we edit the first frame by an image editing model, then use video object dependence for conditional guidance video generation);
generating, using a machine-learning model with inputs of the foreground feature data and the mask data, a composite video that aligns movement of the subject with the second environment, the machine-learning model using the background feature data to generate and condition a depiction of the second environment in the composite video (Wang, page 1, Abstract, video object composition tasks ... ensure that the objects in the composited video maintain motion and identity consistency ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, The proposed multiple video object composition framework is illustrated in Fig. 2 ... the video objects are derived to the corresponding object masks {Myi}N, i=1 by the mask extraction module ... we combine the video objects and the corresponding masks into a new conditional guidance video … we leverage an image editing model to obtain an edited first frame … and feed it into the I2V generation model for video generation … we use the features and attention maps of the corresponding noise {{yit}N, i =1}T, t=1 as condition injections to generate the composited video); and
presenting the composite video via a user interface (Wang, pages 7-8, 5.3. Quantitative results, User study. 25 participants are invited to vote on seven videos composited ... we showed the participants three input videos and two composited videos generated by our method and a compared method, and then recorded the number of participants voting on the harmony and temporal consistency of the composited videos).
As per claim 13, Wang discloses the computing device of claim 11, wherein: the machine-learning model is a convolutional neural network; and the background feature data are injected through cross-attention layers of a denoising U-Net of the convolutional neural network (Wang, page 3, 3.3. Attention mechanisms, The denoising model ϵθ (xt, t) is often instantiated by architectures like UNet, which has four basic modules: convolutional modules, spatial self-attention modules, spatial cross-attention modules, and temporal self-attention modules. In this work, we focus on latent diffusion models [33] for video generation, which formulates the attention operation as: [5a; 5b], where z is the input hidden state to the attention layer and WQ, WK and WV are trainable projection matrices that map z onto query, key and value vectors, respectively; Wang, page 4, Figure 2. Multiple video object composition framework).
As per claim 16, Wang discloses one or more computer-readable storage media storing instructions that, responsive to execution by a processing device, causes the processing device to perform operations comprising:
receive a subject video depicting a movement of a subject in a first environment and a condition frame depicting a second environment, the second environment being different than the first environment (Wang, page 4, Figure 2. Multiple video object composition framework. video object preprocessing and generative video editing. In the preprocessing stage, we perform DDIM inversion, object extraction and paste, as well as mask extraction … we edit the first frame by an image editing model, then use video object dependence for conditional guidance video generation);
generate, using a machine-learning model, a composite video that aligns the movement of the subject with the second environment, inputs to the machine-learning model including mask data of the subject in frames of the subject video, foreground feature data describing latent features of the subject in the frames of the subject video, and background feature data describing latent features of the second environment and being used by the machine-learning model to generate and condition a depiction of the second environment in the composite video (Wang, page 4, Figure 2. Multiple video object composition framework. Our method presents a two-stage approach: video object preprocessing and generative video editing. In the preprocessing stage, we perform … object extraction … as well as mask extraction; Wang, page 1, Abstract, video object composition tasks ... ensure that the objects in the composited video maintain motion and identity consistency ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects; Wang, pages 3-4, 4. The proposed method, 4.1. Overall framework, The proposed multiple video object composition framework is illustrated in Fig. 2 ... the video objects are derived to the corresponding object masks {Myi}N, i=1 by the mask extraction module ... we combine the video objects and the corresponding masks into a new conditional guidance video … we leverage an image editing model to obtain an edited first frame … and feed it into the I2V generation model for video generation … we use the features and attention maps of the corresponding noise {{yit}N, i =1}T, t=1 as condition injections to generate the composited video); and
present the composite video via a user interface (Wang, pages 7-8, 5.3. Quantitative results, User study. 25 participants are invited to vote on seven videos composited ... we showed the participants three input videos and two composited videos generated by our method and a compared method, and then recorded the number of participants voting on the harmony and temporal consistency of the composited videos).
As per claim 19, Wang discloses the one or more computer-readable storage media of claim 16, wherein the condition frame includes a digital photograph of the second environment, a frame of a video depicting the second environment, or a digital image of the second environment generated using another machine-learning model or photo editing resources (Wang, page 4, Figure 2. Multiple video object composition framework. we edit the first frame by an image editing model, then use video object dependence for conditional guidance video generation).
As per claim 20, Wang discloses the one or more computer-readable storage media of claim 16, wherein: the machine-learning model is a convolutional neural network; and the one or more computer-readable storage media store additional instructions that, responsive to execution by a processing device, cause the processing device to perform operations comprising generating, using an image encoder, a feature representation of the condition frame with background feature data being a last hidden layer of the feature representation, the background feature data being injected through cross-attention layers of a denoising U-Net of the convolutional neural network (Wang, page 3, 3.3. Attention mechanisms, The denoising model ϵθ (xt, t) is often instantiated by architectures like UNet, which has four basic modules: convolutional modules, spatial self-attention modules, spatial cross-attention modules, and temporal self-attention modules. In this work, we focus on latent diffusion models [33] for video generation, which formulates the attention operation as: [5a; 5b], where z is the input hidden state to the attention layer and WQ, WK and WV are trainable projection matrices that map z onto query, key and value vectors, respectively; Wang, page 4, Figure 2. Multiple video object composition framework).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 2, 3, 6-8, 12, 14, 17 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang W, Chen Y, Liu Y, Yuan Q, Yang S, Zhang Y. MVOC: a training-free multiple video object composition method with diffusion models. arXiv preprint arXiv:2406.15829. 2024 Jun 22, hereinafter, “Wang” as applied to claims 1, 11 and 16 above, and further in view of Xue Z, Luo M, Chen C, Grauman K. HOI-Swap: Swapping Objects in Videos with Hand-Object Interaction Awareness. arXiv preprint arXiv:2406.07754. 2024 Jun 11, hereinafter, “Xue”.
As per claim 2, Wang disclose he method of claim 1, wherein the machine-learning model is a generative diffusion model (Wang, page 1, Abstract, we propose a Multiple Video Object Composition (MVOC) method based on diffusion models ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects).
Wang does not explicitly disclose the following limitations as further recited however Xue discloses
trained self-supervised on multiple training videos depicting example subject-scene interactions to extrapolate interactions between the subject depicted in the subject video and the second environment depicted in the condition frame into an extended space-time volume in generating the composite video depicting the subject interacting with the second environment (Xue, Abstract, we present HOI-Swap, a novel diffusion-based video editing framework trained in a self-supervised manner. Designed in two stages, the first stage focuses on object swapping in a single frame with HOI awareness; the model learns to adjust the interaction patterns, such as the hand grasp, based on changes in the object’s properties. The second stage extends the single-frame edit across the entire sequence).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Xue with Wang because they are in the same field of endeavor. One skilled in the art would have been motivated to include the self-supervised training as taught by Xue in the system of Wang in order to enable the use of pseudo data pairs using data augmentation for training in lieu of labeled data (Xue, pages 4-5, 3.2. Stage I: HOI-aware Object Swapping in One Frame).
As per claim 3, Wang and Xue disclose the method of claim 2, wherein the generative diffusion model is further trained to infer camera motion from the frames of the subject video in generating the composite video with camera movement within the extended space-time volume of the second environment (Wang, page 9, 6. Limitations, the composited video objects need to keep the same camera motion as the original videos).
As per claim 6, Wang discloses the method of claim 1, and (Wang, page 5, 5. Experiments, 5.1. Experimental setup, We use Segment-and-Track-Anything [10] to extract the masks of moving video objects) but does not explicitly disclose the following limitations as further recited however Xue discloses
wherein the method further comprises: generating, for each frame of the subject video and using an instance segmentation machine-learning model, subject segmentations of the subject and subject masks that localize the subject using a bounding box (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Masked frame preparation: For each frame Ii, we obtain the object’s bounding box (denoted by Mbbox i) from its segmentation mask Mi);
encoding, using a variational autoencoder, the subject segmentations from a pixel space into a latent space as the foreground feature data, the foreground feature data including latent features of the subject (Xue, page 5, Figure 3; Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation); and
downsampling the subject masks into the mask data to align with a size of the foreground feature data (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Object augmentation, At test time, the reference object may appear in any pose, orientation, or size. To bridge this gap between training and testing scenarios, we apply strong augmentation techniques: (1) Spatially, we enhance the object’s diversity by applying random flips, rotations, and perspective transformations ... for preservation of the reference object’s identity and structural details, we collage the augmented reference object image ... onto the masked region).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Xue with Wang because they are in the same field of endeavor. One skilled in the art would have been motivated to substitute the segmentation model as taught by Xue in the system of Wang as an alternate means to obtain the object masks (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame).
As per claim 7, Wang and Xue disclose the method of claim 6, wherein a concatenation of the foreground feature data, the mask data, and Gaussian noises along a feature dimension in the latent space is input to the machine-learning model (Xue, page 5, Figure 3; Xue, pages 5-6, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation ... During the forward diffusion process, Gaussian noise is gradually added to z over T steps, producing a sequence of noisy samples ... We train a denoising network ϵθ1 (◦, t) that learns to reverse the diffusion process, which is implemented as a UNet ... Our LDM is designed to take in two types of conditional signals alongside the standard zt and t: (1) The reference object image is encoded through DINO [3] for distinctive object features dobj ... dobj is then taken by ϵθ1 as input to guide the denoising process via cross-attention mechanisms; (2) The masked frame is encoded by the same VAE to produce zm ... zm is then concatenated channel-wise with z before being fed into ϵθ1).
As per claim 8, Wang and Xue disclose the method of claim 7, wherein: the latent features of the foreground feature data is included in four latent channels; and the Gaussian noises include noisy latent features in the four latent channels (Xue, page 5, Figure 3; Xue, pages 5-6, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation ... During the forward diffusion process, Gaussian noise is gradually added to z over T steps, producing a sequence of noisy samples ... We train a denoising network ϵθ1 (◦, t) that learns to reverse the diffusion process, which is implemented as a UNet ... Our LDM is designed to take in two types of conditional signals alongside the standard zt and t: (1) The reference object image is encoded through DINO [3] for distinctive object features dobj ... dobj is then taken by ϵθ1 as input to guide the denoising process via cross-attention mechanisms; (2) The masked frame is encoded by the same VAE to produce zm ... zm is then concatenated channel-wise with z before being fed into ϵθ1).
As per claim 12, Wang discloses the computing device of claim 11, wherein the machine-learning model is a generative diffusion model Wang, page 1, Abstract, we propose a Multiple Video Object Composition (MVOC) method based on diffusion models ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects).
Wang does not explicitly disclose the following limitations as further recited however Xue discloses
trained self-supervised on multiple training videos depicting example subject-scene interactions to extrapolate interactions between the subject depicted in the subject video and the second environment depicted in the condition frame into an extended space-time volume in generating the composite video depicting the subject interacting with the second environment (Xue, Abstract, we present HOI-Swap, a novel diffusion-based video editing framework trained in a self-supervised manner. Designed in two stages, the first stage focuses on object swapping in a single frame with HOI awareness; the model learns to adjust the interaction patterns, such as the hand grasp, based on changes in the object’s properties. The second stage extends the single-frame edit across the entire sequence).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Xue with Wang because they are in the same field of endeavor. One skilled in the art would have been motivated to include the self-supervised training as taught by Xue in the system of Wang in order to enable the use of pseudo data pairs using data augmentation for training in lieu of labeled data (Xue, pages 4-5, 3.2. Stage I: HOI-aware Object Swapping in One Frame).
As per clam 14, Wang discloses the computing device of claim 13, and (Wang, page 5, 5. Experiments, 5.1. Experimental setup, We use Segment-and-Track-Anything [10] to extract the masks of moving video objects) but does not explicitly disclose the following limitations as further recited however Xue discloses wherein the computer-readable storage medium stores additional instructions that, responsive to execution by the processing device, causes the processing device to perform operations including:
generating, for each frame of the subject video and using an instance segmentation machine-learning model, subject segmentations of the subject and subject masks that localize the subject using a bounding box (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Masked frame preparation: For each frame Ii, we obtain the object’s bounding box (denoted by Mbbox i ) from its segmentation mask Mi);
encoding, using a variational autoencoder, the subject segmentations from a pixel space into a latent space as the foreground feature data, the foreground feature data including latent features of the subject (Xue, page 5, Figure 3; Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation); and
downsampling the subject masks into the mask data to align with a size of the foreground feature data (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Object augmentation, At test time, the reference object may appear in any pose, orientation, or size. To bridge this gap between training and testing scenarios, we apply strong augmentation techniques: (1) Spatially, we enhance the object’s diversity by applying random flips, rotations, and perspective transformations ... for preservation of the reference object’s identity and structural details, we collage the augmented reference object image ... onto the masked region).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Xue with Wang because they are in the same field of endeavor. One skilled in the art would have been motivated to substitute the segmentation model as taught by Xue in the system of Wang as an alternate means to obtain the object masks (Xue, page 5, 3.2 Stage I: HOI-aware Object Swapping in One Frame).
As per claim 17, Wang discloses the one or more computer-readable storage media of claim 16, wherein the machine-learning model is a generative diffusion model Wang, page 1, Abstract, we propose a Multiple Video Object Composition (MVOC) method based on diffusion models ... The final generative model not only constrains the objects in the generated video to be consistent with the original object motion and identity, but also introduces interaction effects between objects).
Wang does not explicitly disclose the following limitations as further recited however Xue discloses
trained self-supervised on multiple training videos depicting example subject-scene interactions to extrapolate interactions between the subject depicted in the subject video and the second environment depicted in the condition frame into an extended space-time volume in generating the composite video depicting the subject interacting with the second environment (Xue, Abstract, we present HOI-Swap, a novel diffusion-based video editing framework trained in a self-supervised manner. Designed in two stages, the first stage focuses on object swapping in a single frame with HOI awareness; the model learns to adjust the interaction patterns, such as the hand grasp, based on changes in the object’s properties. The second stage extends the single-frame edit across the entire sequence).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Xue with Wang because they are in the same field of endeavor. One skilled in the art would have been motivated to include the self-supervised training as taught by Xue in the system of Wang in order to enable the use of pseudo data pairs using data augmentation for training in lieu of labeled data (Xue, pages 4-5, 3.2. Stage I: HOI-aware Object Swapping in One Frame).
As per claim 18, Wang and Xue disclose the one or more computer-readable storage media of claim 17, wherein the generative diffusion model is further trained to infer camera motion from the frames of the subject video in generating the composite video with camera movement within the extended space-time volume of the second environment (Wang, page 9, 6. Limitations, the composited video objects need to keep the same camera motion as the original videos).
Claim(s) 9 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang W, Chen Y, Liu Y, Yuan Q, Yang S, Zhang Y. MVOC: a training-free multiple video object composition method with diffusion models. arXiv preprint arXiv:2406.15829. 2024 Jun 22, hereinafter, “Wang”, in view of Xue Z, Luo M, Chen C, Grauman K. HOI-Swap: Swapping Objects in Videos with Hand-Object Interaction Awareness. arXiv preprint arXiv:2406.07754. 2024 Jun 11, hereinafter, “Xue” as applied to claims 8 and 16 above, and further in view of Esser, Patrick, et al. "Structure and Content-Guided Video Synthesis with Diffusion Models." arXiv preprint arXiv:2302.03011 (2023), hereinafter, “Esser”.
As per claim 9, Wang and Xue disclose the method of claim 8, but do not explicitly disclose the following limitation as further recited however Esser discloses wherein the method further includes: reconstructing, using a decoder, a video output of the machine-learning model in the four latent channels into a pixel space of the composite video (Esser, page 4, 3.1. Latent diffusion models, Latent diffusion, Latent diffusion models (LDMs) take the diffusion process into the latent space. This provides an improved separation between compressive and generative learning phases of the model. Specifically, LDMs use an autoencoder where an encoder E maps input data x to a lower dimensional latent code according to z = E(x) while a decoder D converts latent codes back to the input space ... Our encoder downsamples RGB-images ... by a factor of eight and outputs four channels, resulting in a latent code).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Esser with Wang and Xue because they are in the same field of endeavor. One skilled in the art would have been motivated to include the decoder as taught by Esser in the system of Wang and Xue in order to improve runtime and memory efficiency (Esser, page 4, 3.1. Latent diffusion models, Latent diffusion).
As per claim 15, Wang and Xue disclose the computing device of claim 14, wherein:
a concatenation of the foreground feature data, the mask data, and Gaussian noises along a feature dimension in the latent space is input to the convolutional neural network (Xue, page 5, Figure 3; Xue, pages 5-6, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation ... During the forward diffusion process, Gaussian noise is gradually added to z over T steps, producing a sequence of noisy samples ... We train a denoising network ϵθ1 (◦, t) that learns to reverse the diffusion process, which is implemented as a UNet ... Our LDM is designed to take in two types of conditional signals alongside the standard zt and t: (1) The reference object image is encoded through DINO [3] for distinctive object features dobj ... dobj is then taken by ϵθ1 as input to guide the denoising process via cross-attention mechanisms; (2) The masked frame is encoded by the same VAE to produce zm ... zm is then concatenated channel-wise with z before being fed into ϵθ1);
the latent features of the foreground feature data is included in four latent channels; the Gaussian noises include noisy latent features in the four latent channels (Xue, page 5, Figure 3; Xue, pages 5-6, 3.2 Stage I: HOI-aware Object Swapping in One Frame, Model design, We design the stage-I model as an image-conditioned latent diffusion model (LDM). Specifically, we employ a pretrained variational autoencoder (VAE) E as used in prior work to encode a frame I into a latent space representation ... During the forward diffusion process, Gaussian noise is gradually added to z over T steps, producing a sequence of noisy samples ... We train a denoising network ϵθ1 (◦, t) that learns to reverse the diffusion process, which is implemented as a UNet ... Our LDM is designed to take in two types of conditional signals alongside the standard zt and t: (1) The reference object image is encoded through DINO [3] for distinctive object features dobj ... dobj is then taken by ϵθ1 as input to guide the denoising process via cross-attention mechanisms; (2) The masked frame is encoded by the same VAE to produce zm ... zm is then concatenated channel-wise with z before being fed into ϵθ1).
Wang and Xue do not explicitly disclose the following limitation as further recited however Esser discloses
the computer-readable storage medium stores additional instructions that, responsive to execution by the processing device, causes the processing device to perform operations including reconstructing, using a decoder, a video output of the machine-learning model in the four latent channels into a pixel space of the composite video (Esser, page 4, 3.1. Latent diffusion models, Latent diffusion Latent diffusion models (LDMs) take the diffusion process into the latent space. This provides an improved separation between compressive and generative learning phases of the model. Specifically, LDMs use an autoencoder where an encoder E maps input data x to a lower dimensional latent code according to z = E(x) while a decoder D converts latent codes back to the input space ... Our encoder downsamples RGB-images ... by a factor of eight and outputs four channels, resulting in a latent code).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Esser with Wang and Xue because they are in the same field of endeavor. One skilled in the art would have been motivated to include the decoder as taught by Esser in the system of Wang and Xue in order to improve runtime and memory efficiency (Esser, page 4, 3.1. Latent diffusion models, Latent diffusion).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TRACY MANGIALASCHI whose telephone number is (571)270-5189. The examiner can normally be reached M-F, 9:30AM TO 6:00PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Vu Le can be reached at (571) 272-7332. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/TRACY MANGIALASCHI/Primary Examiner, Art Unit 2668
Prosecution Insights