Abstract: Methods, systems, and apparatuses are described for throttling and management of notifications. A rules manager stores definitions of multiple classes of notifications for sending to recipients. Each of the classes has a respective set of throttling parameters that enables notifications in each class to be selectively throttled. A notification monitor monitors the number of notifications in each class transmitted to a plurality of recipients over time. In response to determining that a threshold number of notifications of a class have been transmitted to a recipient during an interval of time, the recipient is placed in a throttled mode for the class that reduces a subsequent number of notifications transmitted to the recipient. The threshold number may be dependent upon a transmission channel of the notification. The recipient is informed of being placed in the throttled mode and provided an indication of where subsequent notifications can be accessed.

Abstract: Techniques and tools for sub-block transform coding are described. For example, a video encoder adaptively switches between 8×8, 8×4, and 4×8 DCTs when encoding 8×8 prediction residual blocks; a corresponding video decoder switches between 8×8, 8×4, and 4×8 inverse DCTs during decoding. The video encoder may determine the transform sizes as well as switching levels (e.g., frame, macroblock, or block) in a closed loop evaluation of the different transform sizes and switching levels. The encoder and decoder may use different scan patterns for different transform sizes when scanning values from two-dimensional blocks into one-dimensional arrays, or vice versa. The encoder and decoder may use sub-block pattern codes to indicate the presence or absence of information for the sub-blocks of particular blocks.

Abstract: The description relates to activity graphs. One example can obtain details from multiple activity cards. The example can identify relationships between at least some of the details of the multiple activity cards. The example can also generate an activity graph that reflects at least some of the relationships between individual activity cards.

Abstract: Innovations in encoding or decoding when switching color spaces are presented. For example, some of the innovations relate to signaling of control information for adaptive color space transformation (“ACT”). Other innovations relate to ACT operations. These innovations can improve coding efficiency when switching between color spaces during encoding and decoding.

Abstract: An interface, through which functionality of a cloud computing infrastructure can be accessed, can create defined endpoints through which such an interface is accessed, with such defined endpoints limiting the functionality accessible through the interface to only allowed functions. An elevate function can, through a secure key exchange protocol, receive appropriate assurances and can, in response, remove the functionality limitations of the endpoint, thereby enabling unfettered access to the cloud computing infrastructure. Such unrestricted access can be limited in duration, which duration can be established in advance, or agreed-upon through the key exchange mechanism.

Abstract: A series of frames or live images received by a camera can be processed by an automatic image capture system to determine when an automatic image capture should take place. The system can detect when a scene change occurs by comparison of data for two or more frames, and for particular regions in those frames. If a scene change for the specified region is followed by a stabilization of the scene in that region, the system can trigger an auto-capture of at least the stabilized region of the scene, without the need for any user input.

Abstract: A performance investigation tool (PIT) is described herein for investigating the performance of a distributed processing system (DPS). The PIT operates by first receiving input information that describes a graph processing task to be executed using a plurality of computing units. The PIT then determines, based on the input information, at least one time-based performance measure that describes the performance of a DPS that is capable of performing the graphical task. More specifically, the PIT can operate in a manual mode to explore the behavior of a specified DPS, or in an automatic mode to find an optimal DPS from within a search space of candidate DPSs. A configuration system may then be used to construct a selected DPS, using the plurality of computing units. In one case, the graph processing task involves training a deep neural network model having a plurality of layers.

Abstract: A server set may provide a document service to various clients in view of considerations such as availability, fault tolerance, flexibility, and performance. Presented herein are document service architectures that involve partitioning the document set into at least two document ranges, and assigning the respective document ranges to an agent that is deployed to at least one assigned server. A request to apply an operation to a selected document may be fulfilled by identifying the document range of the document; identifying a selected server of the server set that hosts the agent to which the range is assigned; and forwarding the request to the selected server. In some variations, servers may retain detailed information about neighboring servers (e.g., according to logical and/or physical proximity) and scant information about distant servers, thereby avoiding both the extensive information exchange of highly informed network architectures and the inefficiency of uninformed routing algorithms.

Abstract: Online content is served to participant devices using two or more systems. The content served by each system is not the same. Rather, the content streams coming from each system is a partial or lower-quality version of the original high-quality version of the content stream. A single one of the partial data streams can be used by the participant device to output a lower-quality version of the original content stream to the user. Alternately, the received partial content streams can be combined to output, to the user, a high-quality version of the original content stream.

Abstract: Techniques for learned user preference- and behavior-based notification filtering are described herein. In one or more implementations, notifications obtained from computer applications are filtered for presentation to a user. Example notifications include notifications about emails, text messages, phone calls, web-page specific messages, antivirus application messages, and so forth. As part of filtering the notifications, interactions of a user with the notifications and with events for which the notifications can be generated are monitored. The monitored interactions are used to learn user preferences and behaviors for notifications in different contexts of user interaction with computing devices. Data is collected that describes characteristics of a current context. Based on the current context, importance scores are computed for new notifications using the learned user preferences and behaviors. The importance scores can then be used to determine which of the new notifications to present to the user.

Abstract: An input mode trigger is detected so that a computing system treats inputs from a touch sensing device as touch inputs. A focus area input mechanism, which is smaller than a display screen controlled by the computing system, is displayed on the display screen. A hover mode touch input is detected, and a touch input on the touch sensing device is mirrored by corresponding movement of visual indicia on the focus area input mechanism on the display screen. Other touch gestures are used to perform operations within the focus area input mechanism.

Abstract: Systems and methods are provided for evicting entries from a file handle cache. In accordance with certain embodiments, a two-stage eviction process is utilized. In a first stage of the eviction process, entries in the file entry cache are analyzed and marked for eviction while a shared lock is maintained on the file handle cache. The shared lock enables the file handle cache to be concurrently accessed by a content serving system to service content requests. In a second stage of the eviction process, entries in the file handle cache that are marked for eviction are removed while an exclusive lock is maintained on the file handle cache. The exclusive lock prevents the content serving system from concurrently accessing the file handle cache to service content requests.

Abstract: A system for transmitting data over an optical communication path is configured to receive data to be encoded in a bitstream for transmission using an optical communication path and encodes the received data to obtain a bitstream. The system is further configured to determine that the bitstream includes a sequence of consecutive bits, and obtain a power level at which to transmit a portion of the bitstream based on a count of the consecutive bits in the sequence. The system may be configured to selectively activate a light source at a power level according to a modulation scheme to optically transmit the portion of the bitstream at the power level.

Abstract: A presence aggregation system provides a presence aggregation server that allows for the publication of presence state of a publisher from the publisher's multiple endpoints. A presence state includes an availability value and an activity. An activity may include an activity token and/or a custom string. When any one of the publisher's endpoints makes a presence state publication on the presence aggregation server, the presence aggregation server generates an aggregated state of the publisher (i.e., the availability of the publisher aggregated across all of the publisher's endpoints) and publishes the generated aggregated state to each of the publisher's endpoints. The presence aggregation server may also provide the publisher's aggregated state to the subscribers of the publisher's aggregated state information.

Abstract: A preview of dynamic effects in a document is provided. As a user scrolls through pages of a document, for example, slides of a slide presentation, a static representation of each page may be displayed in a main viewing area, and a dynamic view of any dynamic effects may be displayed in an inset preview pane. The preview pane may be provided in a corner of the main viewing area, and may display the dynamic effects in a loop or upon selection. Accordingly, a user may be able to scroll through a document without having to click through interactive content on a page, and concurrently may be informed of and have an opportunity to view the dynamic effects in a preview pane.

Abstract: An overlay network refers to a network that is implemented as various different virtual resources on a physical network referred to as an underlay network. Diagnostics are performed on the overlay network by injecting diagnostic packets from a source endpoint targeting a target endpoint. These endpoints can be in the overlay network, on-premises with the other endpoint but in a different overlay network, or off-premises form the other endpoint. The diagnostic packets include a data packet encapsulated with a diagnostic encapsulation header that can be removed by a network element in the underlay network to allow processing of the data packet, and then added back on. The network element maintains trace information that is a record of receipt of the diagnostic packet and operations performed on the diagnostic packet. A tracing service collects and analyzes this trace information from the various network elements.

Abstract: Methods, systems, apparatuses, and computer program products are provided for a user input device, such as a game controller. The user input device includes a finger depressible trigger button, a trigger travel path adjustment assembly that resides in an internal cavity of a housing of the user input device, and a finger switch configured to be switched between a plurality of selectable positions in real-time (e.g., during game play). A first selectable position of the finger switch enables the trigger button to be depressed a first distance. A second selectable position of the finger switch enables the trigger button to be depressed a second distance that is greater than the first distance.

Abstract: A lens system includes a first lens array assembly including a first plurality of cells, each cell of the first plurality of cells configured to exhibit a pair of first Fourier transform lenses, and a second lens array assembly including a second plurality of cells, each cell of the second plurality of cells configured to exhibit a pair of second Fourier transform lenses. The first and second lens array assemblies are positioned relative to one another along an optical axis of the lens system such that light diverging from an object at a plane disposed at an object conjugate distance from the first lens array assembly reconverges at an image plane after passing through the first and second lens array assemblies. The image plane is disposed at an image conjugate distance from the second lens array assembly in accordance with the object conjugate distance.

Abstract: Non-limiting examples of the present disclosure describe processing operations that achieve file consistency in the presence of a large-scale collaboration service. A mismatch may be determined between hash values associated with two or more versions of a file that is associated with a tenant of a productivity service. Version vector data for different versions of the file may be evaluated. Version vector data may comprise: a session value indicating a session of the productivity service and a version value that indicates a number of changes made by the tenant during the session. A synchronization determination is generated based on an evaluation of the version vector data for the different versions of the file.

Abstract: One or more techniques and/or systems are provided for assigning power management classifications to a process, transitioning a computing environment into a connected standby state based upon power management classifications assigned to processes, and transitioning the computing environment from the connected standby state to an execution state. That is, power management classifications, such as exempt, throttle, and/or suspend, may be assigned to processes based upon various factors, such as whether a process provides desired functionality and/or whether the process provides functionality relied upon for basic operation of the computing environment. In this way, the computing environment may be transitioned into a low power connected standby state that may continue executing desired functionality, while reducing power consumption by suspending and/or throttling other functionality.

Abstract: A system is configured to use a de-randomizer and budget data structure to economize I/O operations for a shared storage device while still allowing access to the device to a number of different entities. Embodiments can identify a comparatively low cost next operation as compared to other I/O operations, including a cost for seek time, for a first entity to dispatch to the storage device when the first entity has sufficient budget to have the I/O operation performed on its behalf and to identify an I/O operation for a second entity to dispatch to the storage device when there is insufficient budget for the first entity.

Abstract: Methods and systems of managing settings are described. According to various embodiments, a settings page corresponding to a user's settings on an online service may be caused to be displayed on a device. The settings page may comprise selectable category setting options for a setting category. The setting category may have a plurality of subcategories, and each subcategory may have its own selectable subcategory setting options. Each one of the category setting options may correspond to its own distinct configuration of subcategory setting options for its subcategories. A user input indicating a selection of one of the category setting options for the setting category may be received. The configuration of subcategory setting options corresponding to the selected one of the category setting options for the setting category may be applied to the user's settings on the online service.

Abstract: In one embodiment, a content aggregation viewer may present a container layout to summarize a data item set to a user. The content aggregation viewer may determine a viewport dimension for a viewport of a user interface of a content aggregation application to present a data item set to a user. The content aggregation viewer may automatically arrange a container set summarizing the data item set in a container layout based on the viewport dimension. The content aggregation viewer may set a parallel container dimension of a data item container of the container set having a variable dimension size in a direction of the viewport dimension based on the viewport dimension. The content aggregation viewer may present the container set in the container layout to the user.

Abstract: The unified system for bimanual interactions provides a lightweight and integrated interface that allows the user to efficiently interact with and manipulate content in the user interface. The system is configured to detect a multi-finger interaction on the touchscreen and to differentiate whether the user intends to pan, zoom or frame a portion of the user interface. Generally, the framing interaction is identified by detection of the user's thumb and forefinger on the touchscreen, which cooperate to define a focus area between vectors extending outwards from the user's thumb and forefinger. Upon a determination that the user intends to interact with or manipulate content within the focus area, the unified system for bimanual interactions provides an indication of the objects that are located within the focus area and contextual menus for interacting with the objects.

Abstract: In some embodiments, the disclosed subject matter involves a server query optimizer for parametric query optimization (PQO) to address the problem of finding and reusing a relatively small number of query plans that can achieve good plan quality across multiple instances of a parameterized query. An embodiment processes query instances on-line and ensures (a) tight, bounded cost sub-optimality for each instance, (b) low optimization overheads, and (c) only a small number of plans need to be stored. A plan re-costing based approach is disclosed to provide good performance on all three metrics. Other embodiments are described and claimed.

Abstract: New methods of relational database management are described, for example, to enable completion and checking of data in relational databases, including completion of missing foreign key values, to facilitate understanding of data in relational databases, to highlight data that it would be useful to add to a relational database and for other applications. In various embodiments, the schema of a relational database is used to automatically create a probabilistic graphical model that has a structure related to the schema. For example, nodes representing individual rows are linked to rows of other tables according to the database schema. In examples, data in the relational database is used to carry out inference using inference algorithms derived from the probabilistic graphical model. In various examples, inference results, comprising probability distributions each for an individual table cell, are used to fill missing data, highlight errors, and for other purposes.

Abstract: Mismatches between schema elements of a data set and a job are identified automatically. Furthermore, the mismatches can be presented visually in conjunction with an interactive visual workspace configured to support diagrammatic authoring of data transformation pipelines. After a data set is connected to a job, one or more mismatches can be determined and presented in context with the workspace. In addition, schema elements can be reconfigured by way of interaction with a visual representation of schema elements to resolve mismatches.

Abstract: Event handler records, for different event handlers in different domains, are stored in an event handler orchestrator service. The event handler records identify event handlers (in various domains) that are to handle events raised in separate domains. When an event is raised, the event handler records are filtered to identify an event handler that has indicated an interest in the raised event, and an end point corresponding to the identified event handler is provided back to the calling process. The calling process then invokes the event handler for which the end point is returned.

Abstract: A search system is provided with an on-line social network system. When a user of the on-line social network system requests a search, the search system examines a member profile representing the user in the on-line social network system and derives, from the member profile, additional terms to be included into the search request together with the user-supplied search terms. The resulting query is processed by applying the AND operator to the user-supplied terms and a weighted OR operator to the additional terms derived from the member profile. The search results obtained by executing the query are made available to the requesting user.

Abstract: A communications link between a computing device and an external device is monitored. A driver for the communications link is executed on the computing device. The driver is configured to monitor data traffic over the communications link. Data indicative of the monitored data traffic is received from the driver. The data is sent to an analysis function configured to identify a condition of the communications link based on accumulated data indicative of the data traffic. Data indicative of an identified condition of the communications link is received from the analytics function. An indication of the identified condition is rendered on a display device of the computing device.

Abstract: Forward and/or backward compatibility between an invoking program and the invoked program can be maintained by using compatibility indicators and/or annotations to configuration settings that specify the desired behavior of the invoked program. Compatibility indicators and/or annotations can be used to override default behavior to ensure compatibility between a version of a program and a version or revision of a dependency of the program. Annotations can be made to configuration settings or to initialization programs and/or files. Configuration and/or initialization settings for the program can be annotated to indicate when default behavior of the dependency is to be overridden to provide alternate behavior. Configuration settings for the program can be annotated to indicate when future revisions to the dependency are to be applied.

Abstract: A technique is described herein for helping a user move a source object that appears on a user interface presentation to a target destination. The technique involves performing semantic analysis on both the source object and at least one candidate target destination. The semantic analysis yields a score that reflects the relationship between the source object and the candidate target destination. The technique then presents a representation of the candidate target destination that is visually enhanced to an extent that is dependent on its output score. The technique then allows the user to move the source object to a representation of a selected target destination, whereupon it becomes associated with the target destination. The technique can produce the enhanced representation by changing a characteristic of an on-screen target identifier (that already exists on the user interface presentation), or by displaying a transient target identifier.

Abstract: A method can include classifying, using a compressed and specialized convolutional neural network (CNN), an object of a video frame into classes, clustering the object based on a distance of a feature vector of the object to a feature vector of a centroid object of the cluster, storing top-k classes, a centroid identification, and a cluster identification, in response to receiving a query for objects of class X from a specific video stream, retrieving image data for each centroid of each cluster that includes the class X as one of the top-k classes, classifying, using a ground truth CNN (GT-CNN), the retrieved image data for each centroid, and for each centroid determined to be classified as a member of the class X providing image data for each object in each cluster associated with the centroid.

Abstract: A peripheral device class identifier is generated for a class of peripheral devices, and is used to identify a setup record that indicates how to install or otherwise set up the class of peripheral devices on a computing device. The peripheral device class identifier is a combination of three components: a vendor identifier, a namespace identifier, and a namespace entry identifier. The vendor identifier is an identifier of the vendor of the class of peripheral devices. The namespace identifier is an identifier of different collections or groups of types of peripheral devices or types of functionality of peripheral devices. The namespace entry identifier is an identifier of a particular type of peripheral device (or particular functionality) of the different collections or groups of types of peripheral devices (or types of functionality of peripheral devices).

Abstract: Features of the present disclosure solve the above-identified problem by implementing remote localization techniques that enable coordination between multi-user display devices. Specifically, the remote localization techniques identify user location (or “position”) in the virtual reality (VR)/augmented reality (AR) scene using a key-frame that is subset of available information. Thus, when a first display device is uncertain regarding its position within a VR/AR scene, the first display may generate, for example, a single key-frame that is shared between one or more second display devices such that the receiving display device(s) may locate the key-frame with respect to a spatial anchor within the client's map, identify the user's location within the VR/AR scene, and transmit the location back to the first HMD device that created the key-frame. Based on the location information, the first HMD device may synchronize with the second HMD device within a shared scene or map.

Abstract: Apparatus and methods are disclosed for allowing smart phone users to “capture the moment” by allowing easy access to a camera application when a mobile device is in an above-lock (or locked) mode, while also preventing unauthorized access to other smart phone functionality. According to one embodiment of the disclosed technology, a method of operating a mobile device having an above-lock state and a below-lock state comprises receiving input data requesting invocation of an camera application when the mobile device is in the above-lock state and invoking the requested camera application on the device, where one or more functions of the requested application are unavailable as a result of the mobile device being in the above-lock state.

Abstract: Systems and methods for providing a predictive framework are provided. The predictive framework comprises plural neural layers of adaptable, executable neurons. Neurons accept one or more input signals and produce an output signal that may be used by an upper-level neural layer. Input signals are received by an encoding neural layer, where there is a 1:1 correspondence between an input signal and an encoding neuron. Input signals for a set of data are received at the encoding layer and processed successively by the plurality of neural layers. An objective function utilizes the output signals of the topmost neural layer to generate predictive results for the data set according to an objective. In one embodiment, the objective is to determine the likelihood of user interaction with regard to a specific item of content in a set of search results, or the likelihood of user interaction with regard to any item of content in a set of search results.

Abstract: An emulator can use compiler metadata to efficiently emulate execution of executable machine code compiled from the source code. Based on accessing compiler metadata associated with machine code, an emulator can identify behavior(s) of the source code from which the machine code is compiled which are not implied by the machine code. From these behaviors, the emulator can identify emulator optimization(s) that can be applied, during emulation of execution of a thread, to reduce a number of steps needed to emulate execution the machine code, while preserving any externally-visible side-effects. These optimizations can operate to reduce a number of emulator operations needed emulate execution of the machine code, or to elide one or more machine code instructions from emulation. These optimizations can then be applied while emulating execution of the thread. The emulated execution could be recorded to a trace that is equivalent to a trace recorded without these optimizations.

Abstract: The mirror deep neural networks (DNNs) as described herein recognize patterns in an input signal. Mirror DNNs regularize to a linear function and train very quickly. Mirror DNNs employ a neural network pattern recognizer that receives a set of features extracted from an input signal and inputs the set of features into a multi-layer neural network. The multi-layer neural network has an input layer that receives the set of features, a plurality of intermediate layers, and an output layer that generates a set of output values that are indicative of a recognized pattern exhibited in the input signal. A first and second non-linear equation pair are chosen and applied to intermediate layers of the neural network so as to make the output values that are indicative of a pattern exhibited in the input signal linear.

Abstract: Saving augmented realities includes collecting, with an augmented reality device, observation information of a physical space, and obtaining, with the augmented reality device, an augmentation associated with the physical space. An augmented view of the physical space including a visual representation of the augmentation is visually presented with the augmented reality device, and the augmented view is saved for subsequent playback.

Abstract: Techniques for graphics processing unit (GPU) partitioning for virtualization are described herein. In one or more implementations, a GPU partitioning manager of a host device obtains a request for a virtual machine having GPU functionality. In particular, the request specifies the GPU functionality in terms of different GPU capabilities. These different capabilities correspond to segments of a GPU model that represents GPU functionality and is used to govern interactions between virtual machines and GPUs. The GPU partitioning manager determines whether GPUs of the host device are available to satisfy the request based on the specified capabilities. If so, the GPU partitioning manager allocates a portion of the determined available GPUs to the virtual machine to configure the virtual machine with a GPU partition having the functionality. The virtual machine configured with the GPU partition can then be exposed to provide GPU-processed data to a GPU partition requestor.

Abstract: The description relates to devices that emit light from a light source. One example device can include a reflector extending along an arc and oriented in light receiving relation relative to the light source. The reflector can have a cross-sectional reflector profile taken transverse to the arc that includes a first portion having a swept profile that transitions to a second portion having a reverse swept profile.

Abstract: Document heading detection includes performing a classification on each of a plurality of paragraphs of a document to identify each paragraph as either a heading or non-heading paragraph. The classification is based on one or more pre-established values corresponding to one or more pre-established formatting features that are indicative of a heading paragraph relative to currently established values for each of the one or more pre-established formatting features in each of the plurality of paragraphs. Document heading detection further includes determining a strength of each of the one or more heading paragraphs by performing a linear regression on each heading paragraph and assigning each of the one or more heading paragraphs a heading level within a hierarchy of heading levels based on the determined strength.

Abstract: The present disclosure provides method and apparatus for identifying flaky tests. Historical running data of a test case may be obtained. Statistical analysis may be performed based on the historical running data. It may be determined whether the test case is a flaky test based on the statistical analysis.

Abstract: A computing device is configured to selectively adjust a color-specific brightness of a video based on a user response to light. For each frame of a plurality of frames included in a video, (1) for a color channel of the frame, a color-specific brightness characterization is determined, (2) for the color channel, in one or more additional frames, one or more additional color-specific brightness characterizations are determined, (3) a filter for the frame is generated based on the color-specific brightness characterization and the one or more additional color-specific brightness characterizations, (4) the filter is applied to the frame to generate a filtered frame, and (5) the filtered frame is output. The filter is configured such that if a color-specific high brightness condition for the color channel is present in the frame, then the color-specific high brightness condition for the color channel is mitigated or removed from the filtered frame.

Abstract: According to implementations of the subject matter described herein, there is proposed a solution for automatic analysis of a difference between multi-dimensional datasets. In this solution, an analysis request is received for a first dataset and a second dataset, each of which including data items corresponding to a plurality of dimensions. In response to the analysis request, data items corresponding to a first dimension in the first and second datasets are compared. Based on the comparison, a first set of influence factors associated with the first dimension are determined, each influence factor indicating a reason for a difference between the first and second datasets from a respective perspective. An analysis result related to the difference between the first and second datasets is presented based on the first set of influence factors. In this way, it is possible to achieve automatic and efficient analysis of the difference between the different datasets.

Abstract: Examples are disclosed herein related to controlling a scanning display system. One example provides a display device comprising a light source, a scanning mirror system configured to scan light from the light source, and a controller configured to control the scanning mirror system to scan the light by synthesizing in a time domain a mirror control waveform that comprises a linear scan portion and a retrace portion stitched to the linear scan portion, the mirror control waveform being continuous and having an arbitrary timing that is adjustable by the controller between scan cycles.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: A method of and system for performing pilot testing of a software program in an organization is carried out by collecting pilot testing data generated from a pilot testing of a software program run on one or more hardware assets in the organization, determining whether a sufficient amount of pilot testing data has been collected, and, when so, calculating one or more pilot test metrics from the collected data. The calculated pilot test metrics may then be compared to similar metrics in a target population to evaluate the software program.

Abstract: A computing device is provided, including a display and a processor. The processor may, for each frame of a plurality of frames included in a video, determine a brightness characterization for a region of interest of the frame. The processor may determine one or more additional brightness characterizations for the region of interest of one or more additional frames of the plurality of frames. The processor may determine that a high brightness condition is present in the region of interest. The processor may generate a filter for the frame based on the brightness characterization and the one or more additional brightness characterizations. The processor may apply the filter to the region of interest of the frame to generate a filtered frame. The high brightness condition is not present in the region of interest of the filtered frame. The processor may output the filtered frame for display at the display.