Patent Application
20190180110
This disclosure relates to a camera system, and more specifically, to processing video data captured using a camera system.
Digital cameras are increasingly used to capture videos in a variety of settings, for instance outdoors or in a sports environment. However, as users capture increasingly more and longer videos, video management becomes increasingly difficult. Manually searching through raw videos (“scrubbing”) to identify the best scenes is extremely time consuming. Automated video processing to identify the best scenes can be very resource-intensive, particularly with high-resolution raw-format video data.
Accordingly, an improved method of automatically identifying the best scenes in captured videos and generating video summaries including the identified best scenes can beneficially improve a user's video editing experience.
The disclosed embodiments have other advantages and features which will be more readily apparent from the following detailed description of the invention and the appended claims, when taken in conjunction with the accompanying drawings, in which:
Figure (or “FIG.”) 1 is a block diagram of a camera system environment according to one embodiment.
The figures and the following description relate to preferred embodiments by way of illustration only. It should be noted that from the following, discussion alternative embodiments of the structures and methods disclosed herein will be readily recognized as viable alternatives that may be employed without departing from the principles of what is claimed.
Reference will now be made in detail to several embodiments, examples of which are illustrated in the accompanying figures. It is noted that wherever practicable similar or like reference numbers may be used in the figures and may indicate similar or like functionality. The figures depict embodiments of the disclosed system (or method) for purposes of illustration only. One skilled in the art will readily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles described herein.
The camera 130 can include a camera body having a camera lens structured on a front surface of the camera body, various indicators on the front of the surface of the camera body (such as LEDs, displays, and the like), various input mechanisms (such as buttons, switches, and touch-screen mechanisms), and electronics (e.g., imaging electronics, power electronics, metadata sensors, etc.) internal to the camera body for capturing images via the camera lens and/or performing other functions. As described in greater detail in conjunction with
The video server 140 receives and stores videos captured by the camera 130 allowing a user to access the videos at a later time. In one embodiment, the video server 140 provides the user with an interface, such as a web page or native application installed on the client device 135, to interact with and/or edit the videos captured by the user. In one embodiment, the video server 140 generates video summaries of various videos stored at the video server, as described in greater detail in conjunction with
Metadata includes information about the video itself, the camera used to capture the video, the environment or setting in which a video is captured or any other information associated with the capture of the video. For example, metadata can include acceleration data representative of the acceleration of a camera 130 attached to a user as the user captures a video while snowboarding down a mountain. Such acceleration metadata helps identify events representing a sudden change in acceleration during the capture of the video, such as a crash the user may encounter or a jump the user performs. Thus, metadata associated with captured video can be used to identify best scenes in a video recorded by a user without relying on image processing techniques or manual curation by a user.
Examples of metadata include: telemetry data (such as motion data, velocity data, and acceleration data) captured by sensors on the camera 130; location information captured by a GPS receiver of the camera 130; compass heading information; altitude information of the camera 130; biometric data such as the heart rate of the user, breathing of the user, eye movement of the user, body movement of the user, and the like; vehicle data such as the velocity or acceleration of the vehicle, the brake pressure of the vehicle, or the rotations per minute (RPM) of the vehicle engine; or environment data such as the weather information associated with the capture of the video. The video server 140 may receive metadata directly from the camera 130 (for instance, in association with receiving video from the camera), from a client device 135 (such as a mobile phone, computer, or vehicle system associated with the capture of video), or from external metadata sources 110 such as web pages, blogs, databases, social networking sites, or servers or devices storing information associated with the user (e.g., a user may use a fitness device recording fitness data).
A user can interact with interfaces provided by the video server 140 via the client device 135. The client device 135 is any computing device capable of receiving user inputs as well as transmitting and/or receiving data via the network 120. In one embodiment, the client device 135 is a conventional computer system, such as a desktop or a laptop computer. Alternatively, the client device 135 may be a device having computer functionality, such as a personal digital assistant (PDA), a mobile telephone, a smartphone or another suitable device. The user can use the client device to view and interact with or edit videos stored on the video server 140. For example, the user can view web pages including video summaries for a set of videos captured by the camera 130 via a web browser on the client device 135.
One or more input devices associated with the client device 135 receive input from the user. For example, the client device 135 can include a touch-sensitive display, a keyboard, a trackpad, a mouse, a voice recognition system, and the like. In some embodiments, the client device 135 can access video data and/or metadata from the camera 130 or one or more metadata sources 110, and can transfer the accessed metadata to the video server 140. For example, the client device may retrieve videos and metadata associated with the videos from the camera via a universal serial bus (USB) cable coupling the camera 130 and the client device 135. The client device can then upload the retrieved videos and metadata to the video server 140.
In one embodiment, the client device 135 executes an application allowing a user of the client device 135 to interact with the video server 140. For example, a user can identify metadata properties using an application executing on the client device 135, and the application can communicate the identified metadata properties selected by a user to the video server 140 to generate and/or customize a video summary. As another example, the client device 135 can execute a web browser configured to allow a user to select video summary properties, which in turn can communicate the selected video summary properties to the video server 140 for use in generating a video summary. In one embodiment, the client device 135 interacts with the video server 140 through an application programming interface (API) running on a native operating system of the client device 135, such as IOS® or ANDROID™. While
The video server 140 communicates with the client device 135, the metadata sources 110, and the camera 130 via the network 120, which may include any combination of local area and/or wide area networks, using both wired and/or wireless communication systems. In one embodiment, the network 120 uses standard communications technologies and/or protocols. In some embodiments, all or some of the communication links of the network 120 may be encrypted using any suitable technique or techniques. It should be noted that in some embodiments, the video server 140 is located within the camera 130 itself.
A controller hub 230 transmits and receives information from various I/O components. In one embodiment, the controller hub 230 interfaces with LED lights 236, a display 232, buttons 234, microphones such as microphones 222, speakers, and the like.
A sensor controller 220 receives image or video input from an image sensor 212. The sensor controller 220 receives audio inputs from one or more microphones, such as microphone 222a and microphone 222b. Metadata sensors 224, such as an accelerometer, a gyroscope, a magnetometer, a global positioning system (GPS) sensor, or an altimeter may be coupled to the sensor controller 220. The metadata sensors 224 each collect data measuring the environment and aspect in which the video is captured. For example, the accelerometer 220 collects motion data, comprising velocity and/or acceleration vectors representative of motion of the camera 130, the gyroscope provides orientation data describing the orientation of the camera 130, the GPS sensor provides GPS coordinates identifying the location of the camera 130, and the altimeter measures the altitude of the camera 130. The metadata sensors 224 are rigidly coupled to the camera 130 such that any motion, orientation or change in location experienced by the camera 130 is also experienced by the metadata sensors 224. The sensor controller 220 synchronizes the various types of data received from the various sensors connected to the sensor controller 220. For example, the sensor controller 220 associates a time stamp representing when the data was captured by each sensor. Thus, using the time stamp, the measurements received from the metadata sensors 224 are correlated with the corresponding video frames captured by the image sensor 212. In one embodiment, the sensor controller begins collecting metadata from the metadata sources when the camera 130 begins recording a video. In one embodiment, the sensor controller 220 or the microcontroller 202 performs operations on the received metadata to generate additional metadata information. For example, the microcontroller may integrate the received acceleration data to determine the velocity profile of the camera 130 during the recording of a video.
Additional components connected to the microcontroller 202 include an I/O port interface 238 and an expansion pack interface 240. The I/O port interface 238 may facilitate the receiving or transmitting video or audio information through an I/O port. Examples of I/O ports or interfaces include USB ports, HDMI ports, Ethernet ports, audioports, and the like. Furthermore, embodiments of the I/O port interface 238 may include wireless ports that can accommodate wireless connections. Examples of wireless ports include Bluetooth, Wireless USB, Near Field Communication (NFC), and the like. The expansion pack interface 240 is configured to interface with camera add-ons and removable expansion packs, such as a display module, an extra battery module, a wireless module, and the like.
Each user of the video server 140 creates a user account, and user account information is stored in the user store 305. A user account includes information provided by the user (such as biographic information, geographic information, and the like) and may also include additional information inferred by the video server 140 (such as information associated with a user's previous use of a camera). Examples of user information include a username, a first and last name, contact information, a user's hometown or geographic region, other location information associated with the user, and the like. The user store 305 may include data describing interactions between a user and videos captured by the user. For example, a user account can include a unique identifier associating videos uploaded by the user with the user's user account.
The video store 310 stores videos captured and uploaded by users of the video server 140. The video server 140 may access videos captured using the camera 130 and store the videos in the video store 310. In one example, the video server 140 may provide the user with an interface executing on the client device 135 that the user may use to upload videos to the video store 310. In one embodiment, the video server 140 indexes videos retrieved from the camera 130 or the client device 135, and stores information associated with the indexed videos in the video store. For example, the video server 140 provides the user with an interface to select one or more index filters used to index videos. Examples of index filters include but are not limited to: the type of equipment used by the user (e.g., ski equipment, mountain bike equipment, etc.), the type of activity being performed by the user while the video was captured (e.g., snowboarding, mountain biking, etc.), the time and data at which the video was captured, or the type of camera 130 used by the user.
In some embodiments, the video server 140 generates a unique identifier for each video stored in the video store 310. In some embodiments, the generated identifier for a particular video is unique to a particular user. For example, each user can be associated with a first unique identifier (such as a 10-digit alphanumeric string), and each video captured by a user is associated with a second unique identifier made up of the first unique identifier associated with the user concatenated with a video identifier (such as an 8-digit alphanumeric string unique to the user). Thus, each video identifier is unique among all videos stored at the video store 310, and can be used to identify the user that captured the video.
The metadata store 325 stores metadata associated with videos stored by the video store 310. For instance, the video server 140 can retrieve metadata from the camera 130, the client device 135, or one or more metadata sources 110, can associate the metadata with the corresponding video (for instance by associating the metadata with the unique video identifier), and can store the metadata in the metadata store 325. The metadata store 325 can store any type of metadata, including but not limited to the types of metadata described herein. It should be noted that in some embodiments, metadata corresponding to a video is stored within a video file itself, and not in a separate storage module.
The web server 330 provides a communicative interface between the video server 140 and other entities of the environment of
The video editing module 320 analyzes metadata associated with a video to identify best scenes of the video based on identified events of interest or activities, and generates a video summary including one or more of the identified best scenes of the video. The video editing module 320 first accesses one or more videos from the video store 310, and accesses metadata associated with the accessed videos from the metadata store 325. The video editing module 320 then analyzes the metadata to identify events of interest in the metadata. Examples of events of interest can include abrupt changes or anomalies in the metadata, such as a peak or valley in metadata maximum or minimum values within the metadata, metadata exceeding or falling below particular thresholds, metadata within a threshold of predetermine values (for instance, within 20 meters of a particular location or within), and the like. The video editing module 320 can identify events of interest in videos based on any other type of metadata, such as a heart rate of a user, orientation information, and the like.
For example, the video editing module 320 can identify any of the following as an event of interest within the metadata: a greater than threshold change in acceleration or velocity within a pre-determined period of time, a maximum or above-threshold velocity or acceleration, a maximum or local maximum altitude, a maximum or above-threshold heart rate or breathing rate of a user, a maximum or above-threshold audio magnitude, a user location within a pre-determined threshold distance from a pre-determined location, a threshold change in or pre-determined orientation of the camera or user, a proximity to another user or location, a time within a threshold of a pre-determined time, a pre-determined environmental condition (such as a particular weather event, a particular temperature, a sporting event, a human gathering, or any other suitable event), or any other event associated with particular metadata.
In some embodiments, a user can manually indicate an event of interest during capture of the video. For example, a user can press a button on the camera, or a camera remote or otherwise interact with the camera during the capture of video to tag the video as including an event of interest. The manually tagged event of interest can be indicated within metadata associated with the captured video. For example, if a user is capturing video while snowboarding and presses a camera button associated with manually tagging an event of interest, the camera creates metadata associated with the captured video indicating that the video includes an event of interest, and indicating a time or portion within the captured video at which the tagged event of interest occurs. In some embodiments, the manual tagging of an event of interest by a user while capturing video is stored as a flag within a resulting video file. The location of the flag within the video file corresponds to a time within the video at which the user manually tags the event of interest.
In some embodiments, a user can manually indicate an event of interest during capture of the video using a spoken command or audio signal. For instance, a user can say “Tag” or “Tag my moment” during the capture of video to tag the video as including an event of interest. The audio-tagged event of interest can be indicated within metadata associated with the captured video. The spoken command can be pre-programmed, for instance by a manufacturer, programmer, or seller of the camera system, or can be customized by a user of the camera system. For instance, a user can speak a command or other audio signal into a camera during a training period (for instance, in response to configuring the camera into a training mode, or in response to the selection of a button or interface option associated with training a camera to receive a spoken command). The spoken command or audio signal can be repeated during the training mode a threshold number of times (such as once, twice, or any number of times necessary for the purposes of identifying audio patterns as described herein), and the camera system can identify an audio pattern associated with the spoken commands or audio signals received during the training period. The audio pattern is then stored at the camera, and, during a video capture configuration, the camera can identify the audio pattern in a spoken command or audio signal received from a user of the camera, and can manually tag an event of interest during the capture of video in response to detecting the stored audio pattern within the received spoken command or audio signal. In some embodiments, the audio pattern is specific to spoken commands or audio signals received from a particular user and can be detected only in spoken commands or audio signals received from the particular user. In other embodiments, the audio pattern can be identified within spoken commands or audio signals received from any user. It should be noted that manually identified events of interest can be associated with captured video by the camera itself, and can be identified by a system to which the captured video is uploaded from the camera without significant additional post-processing.
As noted above, the video editing module 320 can identify events of interest based on activities performed by users when the videos are captured. For example, a jump while snowboarding or a crash while skateboarding can be identified as events of interest. Activities can be identified by the activity identifier module 335 based on metadata associated with the video captured while performing the activities. Continuing with the previous example, metadata associated with a particular altitude and a parabolic upward and then downward velocity can be identified as a “snowboarding jump”, and a sudden slowdown in velocity and accompanying negative acceleration can be identified as a “skateboarding crash”.
The video editing module 320 can identify events of interest based on audio captured in conjunction with the video. In some embodiments, the video editing module identifies events of interest based on one or more spoken words or phrases in captured audio. For example, if audio of a user saying “Holy Smokes!” is captured, the video editing module can determine that an event of interest just took place (e.g., within the previous 5 seconds or other threshold of time), and if audio of a user saying “Oh no! Watch out!” is captured, the video editing module can determine that an event of interest is about to occur (e.g., within the next 5 seconds or other threshold of time). In addition to identifying events of interest based on captured dialogue, the video editing module can identify an event of identify based on captured sound effects, captured audio exceeding a magnitude or pitch threshold, or captured audio satisfying any other suitable criteria.
In some embodiments, the video editing module 320 can identify video that does not include events of interest. For instance, the video editing module 320 can identify video that is associated with metadata patterns determined to not be of interest to a user. Such patterns can include metadata associated with a below-threshold movement, a below-threshold luminosity, a lack of faces or other recognizable objects within the video, audio data that does not include dialogue or other notable sound effects, and the like. In some embodiments, video determined to not include events of interest can be disqualified from consideration for inclusion in a generated video summary, or can be hidden from a user viewing captured video (in order to increase the chance that the remaining video presented to the user does include events of interest).
The activity identifier module 335 can receive a manual identification of an activity within videos from one or more users. In some embodiments, activities can be tagged during the capture of video. For instance, if a user is about to capture video while performing a snowboarding jump, the user can manually tag the video being captured or about to be captured as “snowboarding jump”. In some embodiments, activities can be tagged after the video is captured, for instance during playback of the video. For instance, a user can tag an activity in a video as a skateboarding crash upon playback of the video.
Activity tags in videos can be stored within metadata associated with the videos. For videos stored in the video store 310, the metadata including activity tags associated with the videos is stored in the metadata store 325. In some embodiments, the activity identifier module 335 identifies metadata patterns associated with particular activities and/or activity tags. For instance, metadata associated with several videos tagged with the activity “skydiving” can be analyzed to identify similarities within the metadata, such as a steep increase in acceleration at a high altitude followed by a high velocity at decreasing altitudes. Metadata patterns associated with particular activities are stored in the activity store 340.
In some embodiments, metadata patterns associated with particular activities can include audio data patterns. For instance, particular sound effects, words or phrases of dialogue, or the like can be associated with particular activities. For example, the spoken phrase “nice wave” can be associated with surfing, and the sound of a revving car engine can be associated with driving or racing a vehicle. In some embodiments, metadata patterns used to identify activities can include the use of particular camera mounts associated with the activities in capturing video. For example, a camera can detect that it is coupled to a snowboard mount, and video captured while coupled to the snowboard mount can be associated with the activity of snowboarding.
Once metadata patterns associated with particular activities are identified, the activity identifier module 335 can identify metadata patterns in metadata associated with other videos, and can tag or associate other videos associated with metadata including the identified metadata patterns with the activities associated with the identified metadata patterns. The activity identifier module 335 can identify and store a plurality of metadata patterns associated with a plurality of activities within the activity store 340. Metadata patterns stored in the activity store 340 can be identified within videos captured by one user, and can be used by the activity identifier module 335 to identify activities within videos captured by the user. Alternatively, metadata patterns can be identified within videos captured by a first plurality of users, and can be used by the activity identifier module 335 to identify activities within videos captured by a second plurality of users including at least one user not in the first plurality of users. In some embodiments, the activity identifier module 335 aggregates metadata for a plurality of videos associated with an activity and identifies metadata patterns based on the aggregated metadata. As used herein, “tagging” a video with an activity refers to the association of the video with the activity. Activities tagged in videos can be used as a basis to identify best scenes in videos (as described above), and to select video clips for inclusion in video summary templates (as described below).
Videos tagged with activities can be automatically uploaded to or shared with an external system. For instance, if a user captures video, the activity identifier module 335 can identify a metadata pattern associated with an activity in metadata of the captured video, in real-time (as the video is being captured), or after the video is captured (for instance, after the video is uploaded to the video server 140). The video editing module 320 can select a portion of the captured video based on the identified activity, for instance a threshold amount of time or frames around a video clip or frame associated with the identified activity. The selected video portion can be uploaded or shared to an external system, for instance via the web server 330. The uploading or sharing of video portions can be based on one or more user settings and/or the activity identified. For instance, a user can select one or more activities in advance of capturing video, and captured video portions identified as including the selected activities can be uploaded automatically to an external system, and can be automatically shared via one or more social media outlets.
The video editing module 320 identifies best scenes associated with the identified events of interest for inclusion in a video summary. Each best scene is a video clip, portion, or scene (“video clips” hereinafter), and can be an entire video or a portion of a video. For instance, the video editing module 320 can identify video clips occurring within a threshold amount of time of an identified event of interest (such as 3 seconds before and after the event of interest), within a threshold number of frames of an identified event of interest (such as 24 frames before and after the event of interest), and the like. The amount of length of a best scene can be pre-determined, and/or can be selected by a user.
The amount or length of video clip making up a best scene can vary based on an activity associated with captured video, based on a type or value of metadata associated with captured video, based on characteristics of the captured video, based on a camera mode used to capture the video, or any other suitable characteristic. For example, if an identified event of interest is associated with an above-threshold velocity, the video editing module 320 can identify all or part of the video corresponding to above-threshold velocity metadata as the best scene. In another example, the length of a video clip identified as a best scene can be greater for events of interest associated with maximum altitude values than for events of interest associated with proximity to a pre-determined location.
For events of interest manually tagged by a user, the length of a video clip identified as a best scene can be pre-defined by the user, can be manually selected by the user upon tagging the event of interest, can be longer than automatically-identified events of interest, can be based on a user-selected tagging or video capture mode, and the like. The amount or length of video clips making up best scenes can vary based on the underlying activity represented in captured video. For instance, best scenes associated with events of interest in videos captured while boating can be longer than best scenes associated with events of interest in videos captured while skydiving.
In some embodiments, the video editing module 320 identifies best scenes in a video based on various motion values associated with the video frames of a video. A motion value associated with a frame of a video is a measure of the motion of a portion of or the entire frame of the video. Motion values can include motion vectors, which represent a movement of a frame macroblock from a first frame to a second frame. Similarly, a motion value can represent a motion of an object depicted within the video from a location in a first frame to a location in a second frame (and to subsequent locations in subsequent frames). Motion values can include a first value representative of a motion direction for a video frame portion and a second value representative of motion magnitude of the motion of the video frame portion. A motion value can also be representative of the motion within an entire frame, for instance the sum of all motion vectors associated with portions of the frame. In some embodiments, the video editing module 320 analyzes video to identify motion values for each frame of the video (for instance, for each of a plurality of potions of each frame of the video).
The video editing module may identify portions of the video from which to identify best scenes based on the motion values associated with the frames of the video as is further described in conjunction with
The identified video portions make up the best scenes as described herein. The video editing module 320 generates a video summary by combining or concatenating some or all of the identified best scenes into a single video. The video summary thus includes video portions of events of interest, beneficially resulting in a playable video including scenes likely to be of greatest interest to a user. The video editing module 320 can receive one or more video summary configuration selections from a user, each specifying one or more properties of the video summary (such as a length of a video summary, a number of best scenes for inclusion in the video summary, and the like), and can generate the video summary according to the one or more video summary configuration selections. In some embodiments, the video summary is a renderable or playable video file configured for playback on a viewing device (such as a monitor, a computer, a mobile device, a television, and the like). The video summary can be stored in the video store 310, or can be provided by the video server 140 to an external entity for subsequent playback. Alternatively, the video editing module 320 can serve the video summary from the video server 140 by serving each best scene directly from a corresponding best scene video file stored in the video store 310 without compiling a singular video summary file prior to serving the video summary. It should be noted that the video editing module 320 can apply one or more edits, effects, filters, and the like to one or more best scenes within the video summary, or to the entire video summary during the generation of the video summary.
In some embodiments, the video editing module 320 ranks identified best scenes. For instance, best scenes can be ranked based on activities with which they are associated, based on metadata associated with the best scenes, based on length of the best scenes, based on a user-selected preference for characteristics associated with the best scenes, or based on any other suitable criteria. For example, longer best scenes can be ranked higher than shorter best scenes. Likewise, a user can specify that best scenes associated with above-threshold velocities can be ranked higher than best scenes associated with above-threshold heart rates. In another example, best scenes associated with jumps or crashes can be ranked higher than best scenes associated with sitting down or walking. Generating a video summary can include identifying and including the highest ranked best scenes in the video summary.
In some embodiments, the video editing module 320 classifies scenes by generating a score associated with each of one or more video classes based on metadata patterns associated with the scenes. Classes can include but are not limited to: content-related classes (“snow videos”, “surfing videos”, etc.), video characteristic classes (“high motion videos”, “low light videos”, etc.), video quality classes, mode of capture classes (based on capture mode, mount used, etc.), sensor data classes (“high velocity videos”, “high acceleration videos”, etc.), audio data classes (“human dialogue videos”, “loud videos”, etc.), number of cameras used (“single-camera videos”, “multi-camera videos”, etc.), activity identified within the video, and the like. Scenes can be scored for one or more video classes, the scores can be weighted based on a pre-determined or user-defined class importance scale, and the scenes can be ranked based on the scores generated for the scenes.
In one example, the video editing module 320 analyzes metadata associated with accessed videos chronologically to identify an order of events of interest presented within the video. For example, the video editing module 320 can analyze acceleration data to identify an ordered set of video clips associated with acceleration data exceeding a particular threshold. In some embodiments, the video editing module 320 can identify an ordered set of events occurring within a pre-determined period of time. Each event in the identified set of events can be associated with a best scene; if the identified set of events is chronologically ordered, the video editing module 320 can generate a video summary by a combining video clips associated with each identified event in the order of the ordered set of events.
In some embodiments, the video editing module 320 can generate a video summary for a user using only videos associated with (or captured by) the user. To identify such videos, the video editing module 320 can query the video store 3 I 0 to identify videos associated with the user. In some embodiments, each video captured by all users of the video server I 40 includes a unique identifier identifying the user that captured the video and identifying the video (as described above). In such embodiments, the video editing module 320 queries the video store 310 with an identifier associated with a user to identify videos associated with the user. For example, if all videos associated with User A include a unique identifier that starts with the sequence “X1Y2Z3” (an identifier unique to User A), the video editing module 320 can query the video store 310 using the identifier “X1Y2Z3” to identify all videos associated with User A. The video editing module 320 can then identify best scenes within such videos associated with a user, and can generate a video summary including such best scenes as described herein.
In addition to identifying best scenes, the video editing module 320 can identify one or more video frames that satisfy a set of pre-determined criteria for inclusion in a video summary, or for flagging to a user as candidates for saving as images/photograph stills. The pre-determined criteria can include metadata criteria, including but not limited to: frames with high motion (or blur) in a first portion of a frame and low motion (or blur) in another portion of a frame, frames associated with particular audio data (such as audio data above a particular magnitude threshold or audio data associated with voices or screaming), frames associated with above-threshold acceleration data, or frames associated with metadata that satisfies any other metadata criteria as described herein. In some embodiments, users can specify metadata criteria for use in flagging one or more video frames that satisfy pre-determined criteria. Similarly, in some embodiments, the video editing module 320 can identify metadata patterns or similarities in frames selected by a user to save as images/photograph stills, and can identify subsequent video frames that include the identified metadata patterns or similarities for flagging as candidates to save as images/photograph stills.
In one embodiment, the video editing module 320 retrieves video summary templates from the template store 315 to generate a video summary. The template store 315 includes video summary templates each describing a sequence of video slots for including in a video summary. In one example, each video summary template may be associated with a type of activity performed by the user while capturing video or the equipment used by the user while capturing video. For example, a video summary template for generating video summaries of a ski tip can differ from the video summary template for generating video summaries of a mountain biking trip.
Each slot in a video summary template is a placeholder to be replaced by a video clip or scene when generating a video summary. Each slot in a video summary template can be associated with a pre-defined length, and the slots collectively can vary in length. The slots can be ordered within a template such that once the slots are replaced with video clips, playback of the video summary results in the playback of the video clips in the order of the ordered slots replaced by the video clips. For example, a video summary template may include an introductory slot, an action slot, and a low-activity slot. When generating the video summary using such a template, a video clip can be selected to replace the introductory slot, a video clip of a high-action event can replace the action slot, and a video clip of a low-action event can replace the low-activity slot. It should be noted that different video summary templates can be used to generate video summaries of different lengths or different kinds.
In some embodiments, video summary templates include a sequence of slots associated with a theme or story. For example, a video summary template for a ski trip may include a sequence of slots selected to present the ski trip narratively or thematically. In some embodiments, video summary templates include a sequence of slots selected based on an activity type. For example, a video summary template associated with surfing can include a sequence of slots selected to highlight the activity of surfing.
Each slot in a video summary template can identify characteristics of a video clip to replace the slot within the video summary template, and a video clip can be selected to replace the slot based on the identified characteristics. For example, a slot can identify one or more of the following video clip characteristics: motion data associated with the video clip, altitude information associated with the video clip, location information associated with the video clip, weather information associated with the clip, or any other suitable video characteristic or metadata value or values associated with a video clip. In these embodiments, a video clip having one or more of the characteristics identified by a slot can be selected to replace the slot.
In some embodiments, a video clip can be selected based on a length associated with a slot. For instance, if a video slot specifies a four-second length, a four-second (give or take a pre-determined time range, such as 0.5 seconds) video clip can be selected. In some embodiments, a video clip shorter than the length associated with a slot can be selected, and the selected video clip can replace the slot, reducing the length of time taken by the slot to be equal to the length of the selected video clip. Similarly, a video clip longer than the length associated with a slot can be selected, and either 1) the selected video clip can replace the slot, expanding the length of time associated with the slot to be equal to the length of the selected video clip, or 2) a portion of the selected video clip equal to the length associated with the slot can be selected and used to replace the slot. In some embodiments, the length of time of a video clip can be increased or decreased to match the length associated with a slot by adjusting the frame rate of the video clip to slow down or speed up the video clip, respectively. For example, to increase the amount of time taken by a video clip by 30%, 30% of the frames within the video clip can be duplicated. Likewise, to decrease the amount of time taken by a video clip by 60%, 60% of the frames within the video clip can be removed.
To generate a video summary using a video summary template, the video editing module 320 accesses a video summary template from the template store 315. The accessed video summary template can be selected by a user, can be automatically selected (for instance, based on an activity type or based on characteristics of metadata or video for use in generating the video summary), or can be selected based on any other suitable criteria. The video editing module 320 then selects a video clip for each slot in the video summary template, and inserts the selected video clips into the video summary in the order of the slots within the video summary template.
To select a video clip for each slot, the video editing module 320 can identify a set of candidate video clips for each slot, and can select from the set of candidate video clips (for instance, by selecting the determined best video from the set of candidate video clips according to the principles described above). In some embodiments, selecting a video clip for a video summary template slot identifying a set of video characteristics includes selecting a video clip from a set of candidate video clips that include the identified video characteristics. For example, if a slot identifies a video characteristic of “velocity over 15 mph”, the video editing module 320 can select a video clip associated with metadata indicating that the camera or a user of the camera was traveling at a speed of over 15 miles per hour when the video was captured, and can replace the slot within the video summary template with the selected video clip.
In some embodiments, video summary template slots are replaced by video clips identified as best scenes (as described above). For instance, if a set of candidate video clips are identified for each slot in a video summary template, if one of the candidate video slips identified for a slot is determined to be a best scene, the best scene is selected to replace the slot. In some embodiments, multiple best scenes are identified for a particular slot; in such embodiments, one of the best scenes can be selected for inclusion into the video summary based on characteristics of the best scenes, characteristics of the metadata associated with the best scenes, a ranking of the best scenes, and the like. It should be noted that in some embodiments, if a best scene or other video clip cannot be identified as an above-threshold match for clip requirements associated with a slot, the slot can be removed from the template without replacing the slot with a video clip.
In some embodiments, instead of replacing a video summary template slot with a video clip, an image or frame can be selected and can replace the slot. In some embodiments, an image or frame can be selected that satisfies one or more pre-determined criteria for inclusion in a video summary as described above. In some embodiments, an image or frame can be selected based on one or more criteria specified by the video summary template slot. For example, if a slot specifies one or more characteristics, an image or frame having one or more of the specified characteristics can be selected. In some embodiments, the video summary template slot can specify that an image or frame is to be selected to replace the slot. When an image or frame is selected and used to replace a slot, the image or frame can be displayed for the length of time associated with the slot. For instance, if a slot is associated with a four-second period of display time, an image or frame selected and used to replace the slot can be displayed for the four-second duration.
In some embodiments, when generating a video summary using a video summary template, the video editing module 320 can present a user with a set of candidate video clips for inclusion into one or more video summary template slots, for instance using a video summary generation interface. In such embodiments, the user can presented with a pre-determined number of candidate video clips for a particular slot, and, in response to a selection of a candidate scene by the user, the video editing module 320 can replace the slot with the selected candidate video clip. In some embodiments, the candidate video clips presented to the user for each video summary template slot are the video clips identified as best scenes (as described above). Once a user has selected a video clip for each slot in a video summary template, the video editing module 320 generates a video summary using the user-selected video clips based on the order of slots within the video summary template.
In one embodiment, the video editing module 320 generates video summary templates automatically, and stores the video summary templates in the template store 315. The video summary templates can be generated manually by experts in the field of video creation and video editing. The video editing module 320 may provide a user with a user interface allowing the user to generate video summary templates. Video summary templates can be received from an external source, such as an external template store. Video summary templates can be generated based on video summaries manually created by users, or based on an analysis of popular videos or movies (for instance by including a slot for each scene in a video).
Events of interest within the accessed video are identified 430 based on the accessed metadata associated with the video. Events of interest can be identified based on changes in telemetry or location data within the metadata (such as changes in acceleration or velocity data), based on above-threshold values within the metadata (such as a velocity threshold or altitude threshold), based on local maximum or minimum values within the data (such as a maximum heart rate of a user), based on the proximity between metadata values and other values, or based on any other suitable criteria. Best scenes are identified 440 based on the identified events of interest. For instance, for each event of interest identified within a video, a portion of the video corresponding to the event of interest (such as a threshold amount of time or a threshold number of frames before and after the tittle in the video associated with the event of interest) is identified as a best scene. A video summary is then generated 450 based on the identified best scenes, for instance by concatenating some or all of the best scenes into a single video.
A set of candidate video clips is identified 530 for each slot, for instance, based on the criteria specified by each slot, based on video clips identified as “best scenes” as described above, or based on any other suitable criteria. For each slot, a candidate video clip is selected 540 from among the set of candidate video clips identified for the slot. In some embodiments, the candidate video clips in each set of candidate video clips are ranked, and the most highly ranked candidate video clip is selected. The selected candidate video clips are combined 550 to generate a video summary. For instance, the selected candidate, video clips can be concatenated in the order of the slots of the video summary template with which the selected candidate video clips correspond.
A second video and associated metadata is accessed 740. The metadata pattern is identified 750 within the metadata associated with the second video. Continuing with the previous example, the metadata associated with the second video is analyzed and the defined change in acceleration metadata and altitude metadata is identified within the examined metadata. In response to identifying the metadata pattern within the metadata associated with the second video, the second video is associated 760 with the identified activity.
A motion value associated with each frame of the accessed video is identified 930. In one embodiment, the motion value is determined based on the summation of the magnitude of all the motion vectors in each frame. In one example, each frame is divided into a plurality of portions (e.g., macro blocks) and each portion is associated with a first motion vector along the horizontal axis and a second motion vector along the vertical axis. The motion vectors associated with the portions of a frame represent the motion of elements within the portions of the frame. In this example a motion value is identified 930 for each frame by summing up the motion vectors associated with each portion of the frame. For instance, a net motion vector for each portion is determined by summing the squares of the motion vectors along the horizontal and vertical axis of each portion, and the motion value for the frame is determined by summing the net motion vectors of the portions of the frame. Thus, in this example, the motion value associated with a frame represents a measure of the total or collective motion of the objects in the frame.
A differential motion value is identified 940 for each adjacent pair of frames of the accessed video. As the motion value of a frame represents the velocity of content of the frame, the differential motion value between adjacent frames represents the acceleration of content between frames. In one embodiment, the differential motion value for a pair of frames is identified 940 by determining the difference between the motion value associated with each frame. In one example, a differential motion value is associated with a frame by determining the difference in motion values between a frame and an immediately previous frame.
In some embodiments, a threshold number of differential motion values associated with a sequence of frames of the accessed video are aggregated to identify a single differential motion value for the sequence of frames or an average differential motion value for each pair of frames in the sequence. This helps to reduce possible noise in motion vector values associated with the frames of the accessed video and increases the likelihood of identifying interesting areas of acceleration in the accessed video. For example, motion vectors for macroblocks at corresponding locations for each frame in a group of pictures (or “GOP”) (or, for all frames in the GOP other than the I-frame) can be averaged, producing an array of motion vector averages (one for each macroblock location within a frame) and reducing noise that may otherwise be present within the motion vectors.
Each frame of the accessed video is scored 950 based on the motion value and the differential motion value associated with each frame. The motion value associated with a frame and the differential motion value associated with the frame may be combined in a plurality of different ways to generate a score for the frame. For example, the motion value and the differential motion value may be summed to generate a score for a frame. The score associated with a frame indicates the amount of motion associated with the elements or objects within the frame. Thus, frames with high amounts of motions are associated with higher scores as compared to frames with lower amounts of motion. In some examples, after each frame is scored, the plurality of frames are chronologically ordered and smoothed, with a Gaussian filter for example, to limit the possible undesirable effects of “noisy” video frames.
Best scenes are identified 960 based on the scores associated with the frames of the accessed video. For instance, a set of frames associated with greater than a threshold value of scores are identified and a portion of the accessed video corresponding to each frame in the set of frames (such as a threshold amount of time or a threshold number of frames before and after the frame in the video) is identified as a best scene. In another example, a top number of scores of the plurality of scores (e.g., the top 3 scores) are identified. The corresponding frames associated with top number of scores are identified and portions of the accessed video corresponding to each of the identified frames (such as a threshold amount of time or a threshold number of frames before and after the frame in the video) are identified as best scenes. A video summary is then generated 970 based on the identified best scenes, for instance by concatenating some or all of the best scenes into a single video.
In some embodiments, a set of frames associated scores that are lower than a threshold value (or a threshold number of frames associated with the lowest scores) are identified and a portion of the accessed video corresponding to each frame is identified as a best scene. In some embodiments, sequence of successive frames associated with average scores that are greater than or lower than a pre-determined threshold are identified, and scene is identified based on each identified sequence of frames. In some embodiments, frames are identified by one of motion values associated with the frames and differential motion values associated with the frames, and best scenes are identified based on the identified frames. In some embodiments, frames are identified based on motion values associated with parts of frames. For instance, for frames representative of scenes including the sky, the motion values associated with the portions of frame including the sky may be lower than the motion values associated with the remainder of the frame. In such embodiments, frames can be identified that include a below-threshold motion value associated with a first frame portion and an above-threshold motion value associated with a second frame portion, and scenes can be identified based on these identified frames.
Each video frame of the accessed video is associated with global motion and local motion. The global motion associated with a frame is a measure of the motion of the frame as a whole, while the local motion associated with a frame is a measure of the motion of a portion of the frame with respect to the motion of the other portions of the frame. Thus, a frame associated with a high local motion value includes a portion of the frame with a high motion value with respect to other portions of the frame. The global motion associated with a frame represents a measure of the total motion of all the portions of the frame.
Each video frame is divided 1030 into a number of tiles. The number of tiles each frame is divided into may be based on the resolution of the accessed video. For example, higher resolution video may correspond to a greater number of tiles each frame of the accessed video is divided 1030 into. A motion value associated with each tile of each frame of the accessed video is identified 1040. In one embodiment, the motion value is determined based on the summation of the magnitude of all the motion vectors in each tile.
A motion profile of the accessed video is determined 1050. The motion profile of the accessed video represents the percentage of tiles within each frame having a motion value that is greater than a first threshold value. In alternative embodiments, the motion profile of the accessed video further represents the percentage of tiles within each frame having a motion value below a second threshold value less than the first threshold value. The motion profile of the accessed video may be examined to identify portions of steady global motion, unsteady global motion, and steady local motion in the accessed video. Steady global motion refers to motion associated with a frame or set of frames of the accessed video captured by a substantially motionless camera wherein the motion or activity within the frame or set of frames is substantially discernable to a human viewer or above a motion or activity threshold. Unsteady global motion refers to motion associated with a frame or set of frames of the accessed video captured by a camera in motion wherein the motion or activity within the frame or set of frames is substantially undiscernible to a human viewer or below a motion or activity threshold. Steady local motion refers to local motion within a frame or set of frames of the accessed video captured by a substantially motionless camera that is substantially discernable to a human viewer or above a motion or activity threshold. The motion profile of each accessed video can be determined in advance, or in response to receiving the request to generate a video summary. The motion profile of the accessed video is a measure of global or local motion throughout the accessed video. Portions of the accessed video including steady global or local motion can be identified to determine best scenes for inclusion in a video summary for the accessed video.
In one embodiment, the motion profile of the accessed video is generated by putting together motion profile values associated with each frame of the accessed video. A motion profile value for each frame is determined based on the number of tiles greater than the threshold value, and dividing the number of tiles greater than the threshold value by the total number of tiles within each frame. For example, a frame may include 10 tiles, 5 of which have a motion value greater than the threshold value. The corresponding motion profile value associated with the frame is 0.5 or 50%, as half the tiles are greater than the threshold value. The motion profile for the accessed video can be generated by compiling the motion profile values of each of the frames of the accessed video. Thus, the motion profile for the accessed video can include a sequence of values, wherein each value is associated with a frame and represents the percentage of tiles within the frame that are greater than a first threshold value. In some embodiments, the motion profile for the accessed video can further include a second sequence of values, each associated with a frame and representing the percentage of tiles within the frame that are less than a second threshold value less than the first threshold value.
The motion profile of the accessed video may vary, for instance based on the selection of the first threshold value (and if relevant, second threshold value) and thereby may be examined in a number of ways to identify 1060 best scenes in the accessed video. In one embodiment, the first threshold value is set such that portions of the accessed video including unsteady global motion or fast global motion may be identified. For example, a high threshold value results in portions of the video having motion profile values greater than a predetermined percentage (e.g., 50%) as representing portions of the video with undesirable amounts of fast global motion. Frames with motion profile values greater than the predetermined percentage have a relatively high global motion, and can be classified as unsteady and undesirable, possibly indicative of video captured by a shaky camera. Such frames can be disqualified from consideration as a best scene for inclusion in a video summary prior to the identification of best scenes as described above.
In one embodiment, the second threshold value is set such that portions of the accessed video including slow moving motion, such as steady global or local motion, may be identified. For example, a relatively low second threshold value results in portions of the video having motion profile values lower than a certain percentage (e.g., 50%-70%) as representing portions of the video with slow moving motion or steady global and local motion. Frames with motion profile values lower than the predetermined percentage have a relatively low global and local motion, and can be classified as steady and desirable, possibly indicative of video captured by a steady camera. Best scenes can be identified based on such frames for inclusion in a video summary, as described above.
In one embodiment, the first and/or second threshold values are set such that portions of the accessed video including bands of global motion in the motion profile of the accessed video may be identified. A band of global motion refers to portions of the motion profile of the accessed video including the same or similar motion profile values, such as a first set of frames of the accessed video having a similar motion profile value as a second set of frames. For example, a nominal threshold value results in the motion profile of the accessed video having a plurality of motion bands and a plurality of motion peaks. The motion peaks represent the rapid increases in motion profile values in short periods of time. Given the nominal threshold the motion profile is likely to have motion bands having values greater than a certain percentage as well as motion peaks that rise rapidly above the certain percentage. Motion peaks typically represent rapidly moving undesirable global motion as the percentage of tiles having large motion values rapidly increases between frames in a short period of time—a characteristic that may be attributed to unsteady shots. In some embodiments, motion peaks are scored and weighted based on a width of the motion peak (determined, for instance, to be the full width of the peak at half the maximum amplitude of the peak). In such embodiments, peaks with broader widths (which can represent highlights with consistent motion velocities) may be scored higher than peaks with narrow widths (which can represent short/jerk camera or object motion). Accordingly, best scenes can be selected based on the scored peaks (for instance, scenes associated with above-threshold peak scores can be selected).
Motion bands greater than a certain percentage may represent interesting portions of fast moving steady global motion as a similar percentage of tiles in a set of frames have motion values greater than the nominal threshold value. However, in some examples, motion bands that run for greater than a threshold period of time or include greater than a threshold number of frames with motion values greater than the nominal threshold may represent portions of unsteady global motion that may be undesirable to the user. In this example, the portions of the video including motion bands that last or run for less than a threshold period of time or for less than a threshold number of frames maybe identified 1060 as best scenes. For example, the portions of the accessed video including motion bands may be scored based on one or more of: the motion value associated with each frame, the differential motion value associated with each frame, the length of time of the motion band, whether the motion band exceeds a threshold period of time corresponding to unsteady or undesirable video, and/or the number of frames within the motion band. Accordingly, best scenes may be identified based on the scores, as is further described above.
In one embodiment, the first and/or second threshold values are set such that portions of the accessed video including high amounts of local motion or fast local motion may be identified. For example, a nominal threshold value can be selected such that the motion profile of the accessed video including motion peaks having maximum motion values lower than a certain percentage (e.g., 30%) or within a low range of percentages (e.g., 10%-30%) can be determined. As motion peaks represent portions of the motion profile of the accessed video within which the motion values rise at a rapid pace or within a few number of frames, motion peaks with relatively low motion profile values represent portions of the video including large amounts of local motion, as the percentage of tiles with motion values higher than the threshold value within a frame is relatively low. However, the remaining tiles within the frame and the frames near the frame in question have motion values less than the threshold value, therefore implying that the motion peaks are associated with portions of the accessed video including fast local motion. In this example, the portions of the video including fast local motion or motion peaks having maximum motion values lower than a percentage may be identified 1060 as best scenes. For example, the portions of the accessed video including fast local motion may be scored based on one or more of the motion value associated with each frame, the differential motion value associated with each frame, the amount of fast local motion within each portion, and the number of motion peaks having maximum values lower than a predetermined threshold. Best scenes may then be identified based on the scores, as described above.
A video summary is then generated 1070 based on the identified best scenes, for instance by concatenating some or all of the best scenes into a single video, as described above.
Throughout this specification, some embodiments have used the expression “coupled” along with its derivatives. The term “coupled” as used herein is not necessarily limited to two or more elements being in direct physical or electrical contact. Rather, the term “coupled” may also encompass two or more elements are not in direct contact with each other, but yet still co-operate or interact with each other, or are structured to provide a thermal conduction path between the elements.
Likewise, as used herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having” or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
In addition, use of the “a” or “an” are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give a general sense of the invention. This description should be read to include one or at least one and the singular also includes the plural unless it is obvious that it is meant otherwise.
Finally, as used herein any reference to “one embodiment” or “an embodiment” means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.
Upon reading this disclosure, those of skill in the art will appreciate still additional alternative structural and functional designs for a camera expansion module as disclosed from the principles herein. Thus, while particular embodiments and applications have been illustrated and described, it is to be understood that the disclosed embodiments are not limited to the precise construction and components disclosed herein. Various modifications, changes and variations, which will be apparent to those skilled in the art, may be made in the arrangement, operation and details of the method and apparatus disclosed herein without departing from the spirit and scope defined in the appended claims.
| Number | Date | Country | |
|---|---|---|---|
| 62039849 | Aug 2014 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15467804 | Mar 2017 | US |
| Child | 16256669 | US | |
| Parent | 15091539 | Apr 2016 | US |
| Child | 15467804 | US | |
| Parent | 14705864 | May 2015 | US |
| Child | 15091539 | US |
