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 .
DETAILED ACTION
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.
Claims 1, 3-9, and 11-20 are rejected under 35 U.S.C. 103 as being unpatentable over Palakkode (Pub No. US 20240104693 A1) in view of Xiong (Pub No. US 20230092248 A1).
As per claim 1, Palakkode teaches the claimed:
An apparatus comprising: at least one display; at least one imaging sensor configured to capture image frames of a scene; (Palakkode [0012]: “More specifically, techniques described herein are related to content rendering and AR/VR and multimedia systems by utilizing causal temporal supersampling the video frames using deep learning.” The AR/VR system incudes image capturing and a display.).
Palakkode alone does not explicitly teach the remaining claim limitations.
However, Palakkode in combination with Xiong teaches the claimed:
at least one motion sensor configured to sense motion of the apparatus; (Palakkode teaches tracking motion data for humans involved in the scene. Palakkode [0020]: “According to one or more embodiments, a trained network may be provided to generate synthesized image frames, such as video frames, and virtual of sample and/or reproject the image data rendered at low-resolution GPU rendering 120. In some embodiments, frame synthesis neural network 125 receives, as input signals, one or more image frames, body position parameters associated with the image frames, and/or geometry data for the scene in accordance with the image frames. Body position parameters may include, for example, pose information (e.g., head pose and/or body pose data). In some embodiments, the body position parameters may include additional data, such as motion data, depth data, or other characteristics of the subject of the pose data which may be used for predicting a future pose of the user for a synthesized frame.” The motion of the user is the motion data. The apparatus is an XR device which is used to track a user’s physical motion in relation to the device. Xiong teaches tracking the motion of the XR device. Xiong [0035]: “Also, according to some embodiments, the device 100 may further include a variety of additional resources 180 that can, if permitted, be accessed by the applications 162. According to particular embodiments, the additional resources 180 may include an accelerometer or inertial motion unit 182, which can detect movements of the device 100 along one or more degrees of freedom. As another example, according to particular embodiments, the additional resources 180 may include a dynamic vision sensor (DVS) 184 or one or more cameras 186.” The device is an XR headset. Xiong [0024]: “In some embodiments, the components described with reference to FIG. 1 are components of an AR or XR headset. In other embodiments, the components described with reference to FIG. 1 are components of an accessory device (such as a smartphone) communicatively connected to an AR or XR headset. The embodiment of the device 100 shown in FIG. 1 is for illustration only, and other configurations are possible. Suitable devices come in a wide variety of configurations, and FIG. 1 does not limit the scope of this disclosure to any particular implementation of a device.” The user’s motion is reflected by the motion of the device being worn by the user.).
and at least one processing device configured to:
obtain a first image frame captured at a first time and first depth data associated with the first image frame, the first image frame having a higher resolution than the first depth data; (Palakkode [0012]: “More specifically, techniques described herein are related to content rendering and AR/VR and multimedia systems by utilizing causal temporal supersampling the video frames using deep learning. According to some embodiments, image frames may be captured of the scene at a first frame rate. A system may obtain signals related to upsampling and reprojection in order to synthesize future frames related to the captured image frames. This may include, for example, pose data, gaze data, depth data, geometry data and the like, or some combination thereof. The input signals and the captured image frames may be applied to a trained network which is configured to generate synthesized frames based on a predicted updated field-of-view, and at a resolution which may differ from the resolution of the image frames as captured. The trained network may be configured to consider temporal data, for example in the form of a series of frames used as input. Time series samples may be obtained from each input signal to allow the network to predict one frame, or a series of future frames, based on given input data. Further, in some embodiments, the obtain frames may be rendered at a low frame rate and/or resolution. The network may be capable of generating the synthesized frames at an upsampled target resolution, for example, because some of the input signals may be received at a higher target frame rate and/or resolution. Accordingly, a graphics processor may render a set of frames at a low frame rate, and additional frames, or intermediate frames, may be generated at a higher frame rate or resolution, for example by a neural engine or other component configured to execute a trained model. Accordingly, techniques described herein solve the technical problem of producing high resolution video frames, or frames at a high frame rate on power constrained devices.” There are different frames at different times with corresponding depth data. This includes an initial one.).
predict motion of the apparatus between the first time and a second time; (Palakkode [0026]: “The flowchart 200 continues at block 210, where body position parameters are determined in association with the one or more frames. As described above, body position parameters may include, for example, pose information (e.g., head pose and/or body pose data). In some embodiments, the body position parameters may include additional data, such as motion data, depth data, or other characteristics of the subject of the pose data which may be used for predicting a future pose of the user for a synthesized frame. In some embodiments, the pose data may include data related to the position and orientation of the tracked subject, for example as represented in the form of 6 degrees of freedom. Other characteristics may be obtained, for example, from visual inertial odometry (VIO) tracking, and/or simultaneous localization mapping (SLAM) tracking techniques. In addition to pose information, body position parameters may also include gaze information. Gaze information may include, for example, gaze vector information, pupil location, or the like. Gaze information may be obtained from a gaze tracking pipeline, and may include data collected from gaze tracking sensors, such as cameras facing a user's eye, or other such sensors.” A predicted future pose implies predicted motion between a first and second time.).
generate second depth data based on the first depth data, the first image frame, and the predicted motion, the second depth data having a higher resolution than the first depth data; (Palakkode [0012]: “More specifically, techniques described herein are related to content rendering and AR/VR and multimedia systems by utilizing causal temporal supersampling the video frames using deep learning. According to some embodiments, image frames may be captured of the scene at a first frame rate. A system may obtain signals related to upsampling and reprojection in order to synthesize future frames related to the captured image frames. This may include, for example, pose data, gaze data, depth data, geometry data and the like, or some combination thereof. The network may be capable of generating the synthesized frames at an upsampled target resolution, for example, because some of the input signals may be received at a higher target frame rate and/or resolution. Accordingly, a graphics processor may render a set of frames at a low frame rate, and additional frames, or intermediate frames, may be generated at a higher frame rate or resolution, for example by a neural engine or other component configured to execute a trained model. Accordingly, techniques described herein solve the technical problem of producing high resolution video frames, or frames at a high frame rate on power constrained devices.” The generated synthesized frames include depth data, which is the second depth data. The updated frame would have updated resolution. Xiong teaches outputting new depth maps based on previous depth data. Xiong [0051]: “Referring to the illustrative example of FIG. 4, the output of the first neural network (such as a depth map) is provided to an image-guided super-resolution stage 405, which refines and increases the data density of the depth map output by the first neural network 401 based at least in part on the image data 205.”).
reproject the first image frame using the second depth data to generate a second image frame; (Palakkode [0011]: “Embodiments described herein relate to a technique for generating synthesized images to provide spatial and temporal supersampling. In addition, embodiments described herein are directed to a technique for generating synthesized images in which the field-of-view is reprojected.” As taught above, the original images include depth data, and the synthesized reprojected image which is based on the newly generated, supersampled depth data data.).
and initiate presentation of a rendered image based on the second image frame, the at least one display configured to present the rendered image substantially at the second time; (Palakkode [0054]: “In one or more embodiments, the frame synthesis neural network 430 may be trained to generate synthesized frames based on time series sample data and current data based on sets of rendered frames. For example, a set of frames may be rendered, and contextual information for the frames may be obtained, such as, for each frame, pose data, gaze data, and/or geometry data. A subset of the rendered frames may fed into the frame synthesis network 430 during a training stage, along with the pose data, gaze data, and geometry data. The frame synthesis neural network 430 may predict one or more additional frames which were not included in the input data. The frame synthesis neural network 430 may include an error function which can then compare the predicted frame to the provided frame from the set of rendered frames to determine an error” Rendering the time series means rendering at the second frame.).
wherein, to generate the second depth data, the at least one processing device is configured to perform depth densification and super-resolution in order to increase the resolution of the second depth data relative to the resolution of the first depth data. (Xiong [0075]: “According to some embodiments, at an operation 920, the image data and the sparse depth data are passed to a second neural network (such as the first neural network 401 in FIG. 4) to obtain one or more dense depth maps, where each dense depth map is associated with an ROI for which an object ROI and a feature map ROI were generated. In some embodiments, the operation 920 includes passing the image and sparse depth data to a neural network to obtain an initial set of predicted depth values and performing image-guided super resolution (such as by applying the method 500 of FIG. 5) to increase the resolution (densify) the sparse depth map.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the densification and increased resolution as taught by Xiong with the system of Palakkode in order to increase the data points and level of detail of previously existing depth data to analyze it at a higher resolution through interpolating new data within the existing data.
As per claims 9 and 17, these claims are similar in scope to limitations recited in claim 1, and thus are rejected under the same rationale. As per claim 17, Palakkode teaches a non-transitory computer-readable medium. Palakkode [0063]: “Electronic device 600 may also include storage 650. Storage 650 may include one or more non-transitory computer-readable mediums including, for example, magnetic disks (fixed, floppy, and removable) and tape, optical media such as CD-ROMs and digital video disks (DVDs), and semiconductor memory devices such as Electrically Programmable Read-Only Memory (EPROM) and Electrically Erasable Programmable Read-Only Memory (EEPROM). Storage 650 may be utilized to store various data and structures which may be utilized for presenting virtual objects in an XR environment.”
As per claim 3, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Xiong teaches the claimed:
3. The apparatus of Claim 1, wherein: the at least one processing device is further configured to generate a feature map based on the first image frame; (Xiong [0054]: “Referring to the illustrative example of FIG. 5, the method 500 takes, as inputs, the one or more frames of image data 205 and a low-resolution depth map 505. In some embodiments, the low-resolution depth map 505 may be output from a neural network trained to estimate depth values from sparse depth and image data (such as the first neural network 401 in FIG. 4). In other embodiments, the low-resolution depth map 505 may be obtained elsewhere. Using the image data 205 and low-resolution depth map 505, the method 500 outputs one or more dense depth maps 260 covering the one or more ROIs extracted by the detection and extraction network 215…” This is used to generate the feature map.).
and the at least one processing device is configured to use the feature map during depth densification and super-resolution. (Xiong [0075]: “According to some embodiments, at an operation 920, the image data and the sparse depth data are passed to a second neural network (such as the first neural network 401 in FIG. 4) to obtain one or more dense depth maps, where each dense depth map is associated with an ROI for which an object ROI and a feature map ROI were generated. In some embodiments, the operation 920 includes passing the image and sparse depth data to a neural network to obtain an initial set of predicted depth values and performing image-guided super resolution (such as by applying the method 500 of FIG. 5) to increase the resolution (densify) the sparse depth map.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the feature map generation as taught by Xiong with the system of Palakkode in order to determine areas of interest and characteristics of the image to guide the densification and increase of resolution. The motivation of claim 1 is incorporated herein.
As per claim 11, this claim is similar in scope to limitations recited in claim 3, and thus is rejected under the same rationale.
As per claim 4, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Xiong teaches the claimed:
4. The apparatus of Claim 3, wherein, during depth densification and super- resolution, the at least one processing device is configured to:
map first depth values of the first depth data onto a first set of points; (Xiong [0056]: “According to various embodiments, at an operation 515, a depth value for each pixel or coordinate location of a dense depth map is computed based a function of a weighted average of depth values of neighboring depth points of the sparse depth map. According to some embodiments, computing the depths at the operation 515 includes passing a Gaussian filter over the low-resolution depth map 505 to determine the weighted average value depths for points of the high-resolution depth map 260. In embodiments in which the image data 205 and the low-resolution depth map 505 are provided for only the identified ROIs of a source image frame, the operations 510 and 515 may be looped and performed individually to obtain a separate high-resolution depth map 260 for each ROI.” The points in the sparse map are the original points and may have empty data points between them.).
generate second depth values; and map the second depth values onto a second set of points such that the first depth values and the second depth values together form at least part of the second depth data. (Xiong [0056]: “According to various embodiments, at an operation 515, a depth value for each pixel or coordinate location of a dense depth map is computed based a function of a weighted average of depth values of neighboring depth points of the sparse depth map. According to some embodiments, computing the depths at the operation 515 includes passing a Gaussian filter over the low-resolution depth map 505 to determine the weighted average value depths for points of the high-resolution depth map 260. In embodiments in which the image data 205 and the low-resolution depth map 505 are provided for only the identified ROIs of a source image frame, the operations 510 and 515 may be looped and performed individually to obtain a separate high-resolution depth map 260 for each ROI.” The depth values of the sparse first map remain in corresponding coordinates along with the additional data points of the dense map.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the mapping of depth values specific points as taught by Xiong with the system of Palakkode in order to tie the original and generated depth data to specific regions of the image and to compare the two. The motivation of claim 1 is incorporated herein.
As per claim 12, this claim is similar in scope to limitations recited in claim 4, and thus is rejected under the same rationale.
As per claim 5, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Xiong teaches the claimed:
5. The apparatus of Claim 4, wherein: the at least one processing device is configured to perform depth densification to generate additional depth values not included among the first depth values of the first depth data; (Xiong [0060]: “Referring to the non-limiting example of FIG. 6, the method 600 takes, as inputs, the image data 205, coordinate values specifying the location(s) of the ROI(s) of one or more detected objects 240, the feature map 250 for each ROI, and the dense depth map 260 for each ROI. In the architecture 200, the detection and extraction network 215 outputs one or more detected objects 240 (such as one or more bounding boxes or other definitions of the space in the image data containing the detected objects) based on the outputs of a plurality of multi-scale residual blocks. Thus, ROIs for detected objects may be of different scales. Additionally, the depth map densification network 230 may output depth maps of different size(s) than the source data. Accordingly, there is no expectation or requirement that the ROI(s) of the detected object(s) 240 and the associated feature and dense depth maps will be of equal scale or include an equivalent number of pixels or data points.” Densification involves filling out the data with additional values that were not in the original set.).
and the at least one processing device is configured to perform depth super-resolution to upscale the first depth values and the additional depth values in order to generate the second depth values. (Xiong [0076]: “At an operation 925, the object ROIs, feature map ROIs, and depth map ROIs are aligned and up- or down-scaled as necessary to create input patches of equal size (having the same number of data points) for a fully convolutional segmentation network. In some embodiments, where alignment requires increasing the resolution of one or more of an object, feature map, or depth map ROI, the same image-guided super-resolution techniques used to densify the one or more sparse depth maps at the operation 920 may be applied for upsampling the ROI.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the generation of data not in the original data set through densification as taught by Xiong with the system of Palakkode in order to generate a higher resolution depth map or a depth map with more detailed features. The motivation of claim 1 is incorporated herein.
As per claims 13 and 19, these claims are similar in scope to limitations recited in claim 5, and thus are rejected under the same rationale.
As per claim 6, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Xiong teaches the claimed:
6. The apparatus of Claim 5, wherein the at least one processing device is configured to use a depth filter to generate the additional depth values based on (i) neighboring first depth values of the first depth data, (ii) information from the first image frame, and (iii) the feature map. (Xiong [0042]: “In preparation for segmentation by a convolutional network using a specified image patch size, an alignment operation 270 is performed to align the image data of the object ROI(s), feature map(s), and dense depth map(s) and to resize at least some of the object ROI(s), feature map(s), and dense depth map(s) to a common size scale. According to various embodiments, rescaling to align image data, feature maps, and dense depth maps may also be performed with an image-guided filter, similar to the image-guided filter used to perform depth map densification. At an operation 280, the aligned image data ROI(s), feature map(s), and depth map(s) from the region(s) of interest are passed to a fully convolutional network to obtain the segmentation results 299.” Xiong [0056]: “According to various embodiments, at an operation 515, a depth value for each pixel or coordinate location of a dense depth map is computed based a function of a weighted average of depth values of neighboring depth points of the sparse depth map. According to some embodiments, computing the depths at the operation 515 includes passing a Gaussian filter over the low-resolution depth map 505 to determine the weighted average value depths for points of the high-resolution depth map 260. In embodiments in which the image data 205 and the low-resolution depth map 505 are provided for only the identified ROIs of a source image frame, the operations 510 and 515 may be looped and performed individually to obtain a separate high-resolution depth map 260 for each ROI.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the adjacent original depth data as taught by Xiong with the system of Palakkode in order to interpolate depth data to fill out detail and increase resolution for generated depth data. The motivation of claim 1 is incorporated herein.
As per claims 14 and 20, these claims are similar in scope to limitations recited in claim 6, and thus are rejected under the same rationale.
As per claim 7, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Xiong teaches the claimed:
7. The apparatus of Claim 3, wherein the at least one processing device is configured to perform depth densification using (i) image feature information from the feature map and (ii) at least one of: spatial information, image color texture information, or temporal information from the first image frame. (Xiong [0075]: “According to some embodiments, at an operation 920, the image data and the sparse depth data are passed to a second neural network (such as the first neural network 401 in FIG. 4) to obtain one or more dense depth maps, where each dense depth map is associated with an ROI for which an object ROI and a feature map ROI were generated. In some embodiments, the operation 920 includes passing the image and sparse depth data to a neural network to obtain an initial set of predicted depth values and performing image-guided super resolution (such as by applying the method 500 of FIG. 5) to increase the resolution (densify) the sparse depth map.” Xiong [0069]: “FIG. 8B provides an illustrative visualization of a semantic segmentation 805 of the frame of image data 800 from FIG. 8A. Referring to the illustrative example of FIG. 8B, each the constituent pixels of the image data 800 has been classified (such as by the fully convolutional network 700 in FIG. 7 or another a machine learning tool trained for semantic segmentation, like DeepLab) and shaded, colored, or otherwise marked according to their classifications in a semantic segmentation mask. In this example, pixels associated with human forms (labeled “person pixels” in the figure) have been colored white, while background components of the scene have been colored in shades of dark grey.” Palakkode [0011]: “Embodiments described herein relate to a technique for generating synthesized images to provide spatial and temporal supersampling. In addition, embodiments described herein are directed to a technique for generating synthesized images in which the field-of-view is reprojected.” The generation of synthesized images includes generation of depth data, which is based on temporal information.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the feature and color information for data densification as taught by Xiong with the system of Palakkode in order to generate data based on the important characteristics of the image. The motivation of claim 1 is incorporated herein.
As per claim 15, this claim is similar in scope to limitations recited in claim 7, and thus is rejected under the same rationale.
As per claim 8, Palakkode teaches the claimed:
8. The apparatus of Claim 1, wherein, to perform depth densification and super- resolution, the at least one processing device is configured to use at least one of image correspondence or image feature correspondence between the first image frame and a third image frame, the first and third image frames representing left and right image frames of a stereo pair of image frames. (Palakkode [0025]: “The flowchart 200 begins at block 205, where frames of the scene are captured in a first frame rate. According to one or more embodiments, the frames of a scene may be captured by one or more cameras of an electronic device. In some embodiments, the cameras may be comprised in a wearable device donned by a user. As described above, in some embodiments, frames may be captured of the scene at a low frame rate, or at least rendered at a low frame rate, to conserve power resources of the GPU. Subsequently, a machine learning network may be utilized to generate synthesized frames according to a target frame rate and resolution. The frames of the scene may be captured by one or more cameras of an electronic device. For example, in some embodiments, the frames of the scene may be captured by a stereoscopic or other multi-camera system. As such, in some embodiments, the frames of the scene may include a left frame and a right frame, or may otherwise include multiple frames for a particular capture time. For example, the electronic device may be configured such that the camera is capturing the frames may be synchronized in order to ensure that the content of the frames are captured concurrently.”).
As per claim 16, this claim is similar in scope to limitations recited in claim 8, and thus is rejected under the same rationale.
Claims 2 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Palakkode in view of Xiong and further in view of Yao (Pub No. US 11900622 B2).
As per claim 2, Palakkode alone does not explicitly teach the claimed limitations.
However, Palakkode in combination with Yao teaches the claimed:
2. The apparatus of Claim 1, wherein the resolution of the second depth data matches or substantially matches the resolution of the first image frame. (Yao col. 1 lines 35-45: “Furthermore, there is also known a technique for generating dense depth information by using a stereo image and sparse depth information and to minimize a cost function (see, for example, Non Patent Literature 2). In this technique, dense depth information that reduces an error between the dense depth information and a parallax of the stereo image and an error between the dense depth information and the measured sparse depth information is derived while maintaining depth continuity.” The generation corresponds to the second depth data based on an image. Yao col. 2 lines 28-44: “An aspect of the present invention is a depth hyper-resolving apparatus including: an input data processing unit configured to output a hierarchical input image and hierarchical input depth information by resolution conversion in accordance with a predetermined number of tiers for an input image and input depth information: a depth continuity estimation unit configured to derive a hierarchical estimated depth continuity based on the hierarchical input image: a depth continuity mask deriving unit configured to output, based on the hierarchical input image and the hierarchical estimated depth continuity, a hierarchical depth continuity mask representing values of locations depending on whether a depth is continuous; and a cost function minimization unit configured to derive hyper-resolved depth information to minimize a cost function expressed by using the hierarchical input depth information, the hierarchical depth continuity mask, and the hyper-resolved depth information.”
The estimated depth continuity corresponds to the densified second depth data. Yao col. 5 lines 39-59: “In step S102, the depth continuity estimation unit 24 derives the hierarchical estimated depth continuity 48 using the hierarchical input image 44 as input. Here, the estimated depth continuity is information having the same resolution as the image, and the derivative of the value correlates with the derivative of the depth. For example, the estimated depth continuity may be obtained using a DNN that estimates depth information from an image. As in Non Patent Literature 3 and Non Patent Literature 4, the DNN that estimates depth information can learn only from images, so that training data is easily obtained. Specifically, the DNN that estimates depth information from a monocular image learns using stereo images…”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the generated map that is the same resolution as the original image as taught by Yao with the system of Palakkode in order to allow a map with the same image resolution to be densified to fill in empty data points. The motivation of claim 1 is incorporated herein.
As per claim 10, this claim is similar in scope to limitations recited in claim 2, and thus are rejected under the same rationale.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to THOMAS JOHN FOSTER whose telephone number is (571)272-5053. The examiner can normally be reached Mon, Fri 8:30-6. Tues-Thurs 7:30-5.
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, Daniel Hajnik can be reached at 571-272-7642. 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.
/THOMAS JOHN FOSTER/Examiner, Art Unit 2616
/HAI TAO SUN/Primary Examiner, Art Unit 2616
Prosecution Insights