Efficient Test-Time Adaptation for Improved Temporal Consistency in Video Processing

Abstract

A method for processing a video includes receiving a video as an input at a first layer of an artificial neural network (ANN). A first frame of the video is processed to generate a first label. The artificial neural network is then updated based on the first label. The updating is performed while concurrently processing a second frame of the video. In doing so, temporal inconsistencies between the labels are reduced.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. patent application Ser. No. 17 / 198,147, filed March 10, 2021, entitled "EFFICIENT TEST-TIME ADAPTATION FOR IMPROVED TEMPORAL CONSISTENCY IN VIDEO PROCESSING," the entire disclosure of which is expressly incorporated by reference into this specification.

[0002] Aspects of the present disclosure relate generally to neural networks, and more particularly, to video processing. [Background technology]

[0003]

[0003] An artificial neural network may comprise an interconnected group of artificial neurons (e.g., neuron models). An artificial neural network may be a computational device or may be represented as a method to be implemented by a computational device.

[0004]

[0004] A neural network consists of operands that consume tensors and produce tensors. Neural networks can be used to solve complex problems, but the network size and the number of calculations that can be performed to produce a solution can be large, so the time it takes for the network to complete the task can be long. Furthermore, the computational cost of deep neural networks can be an issue because these tasks can be performed on mobile devices that may have limited computational power.

[0005]

[0005] A convolutional neural network is a type of feed-forward artificial neural network. A convolutional neural network may include a collection of neurons, each with a receptive field, that collectively tile the input space. Convolutional neural networks (CNNs), such as deep convolutional neural networks (DCNs), have numerous applications. In particular, these neural network architectures are used in a variety of technologies, such as image recognition, pattern recognition, speech recognition, autonomous driving, object segmentation in video streams, video processing and other classification tasks.

[0006]

[0006] Modern deep learning-based video processing methods or models may generate inconsistent outputs over time. In some cases, the inconsistency may be observed in the form of flickering or other inconsistencies in the display. Temporally inconsistent outputs (e.g., flickering) may degrade the user's experience and enjoyment, as well as system stability and performance. Summary of the Invention

[0007]

[0007] In one aspect of the disclosure, a method for processing a video is provided. The method includes receiving the video as an input at a first layer of an artificial neural network (ANN). The method also includes processing a first frame of the video to generate a first label. Further, the method includes updating the artificial neural network based on the first label. The updating of the artificial neural network is performed while concurrently processing a second frame of the video.

[0008] In another aspect of the disclosure, an apparatus for processing a video is provided. The apparatus includes a memory and one or more processors coupled to the memory. The processor(s) are configured to receive the video as an input in a first layer of an artificial neural network (ANN). The processor(s) are also configured to process a first frame of the video to generate a first label. Furthermore, the processor(s) are configured to update the artificial neural network based on the first label. The update of the artificial neural network is performed while concurrently processing a second frame of the video.

[0009]

[0009] In one aspect of the disclosure, an apparatus for processing a video is provided. The apparatus includes means for receiving a video as an input at a first layer of an artificial neural network. The apparatus also includes means for processing a first frame of the video to generate a first label. Additionally, the apparatus includes means for updating the artificial neural network based on the first label. The updating of the artificial neural network is performed while concurrently processing a second frame of the video.

[0010] In a further aspect of the present disclosure, a non-transitory computer readable medium is provided. The computer readable medium has encoded thereon program code for processing a video. The program code, when executed by a processor, includes code for receiving the video as an input at a first layer of an artificial neural network. The program code also includes code for processing a first frame of the video to generate a first label. Additionally, the program code includes code for updating the artificial neural network based on the first label. The updating of the artificial neural network is performed while concurrently processing a second frame of the video.

[0011] In one aspect of the disclosure, a method for processing a video is provided. The method includes receiving a video as an input at a first layer of a first artificial neural network and a second artificial neural network. The first artificial neural network has a smaller number of channels than the second artificial neural network. The method also includes processing a first frame of the video through the first artificial neural network to generate a first label. The first artificial neural network provides intermediate features extracted from the first frame of the video to the second artificial neural network. Further, the method includes processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame. Further, the method includes updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video.

[0012]

[0012] In one aspect of the disclosure, an apparatus for processing a video is provided. The apparatus includes a memory and one or more processors coupled to the memory. The processor(s) is configured to receive the video as an input at a first layer of a first artificial neural network and a second artificial neural network. The first artificial neural network has fewer channels than the second artificial neural network. The processor(s) is also configured to process a first frame of the video through the first artificial neural network to generate a first label. The first artificial neural network supplies intermediate features extracted from the first frame of the video to the second artificial neural network. Furthermore, the processor(s) is configured to process the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame. Furthermore, the processor(s) is configured to update the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video.

[0013]

[0013] In one aspect of the disclosure, an apparatus for processing a video is provided. The apparatus includes means for receiving a video as an input at a first layer of a first artificial neural network and a second artificial neural network. The first artificial neural network has fewer channels than the second artificial neural network. The apparatus also includes means for processing a first frame of the video through the first artificial neural network to generate a first label. The first artificial neural network supplies intermediate features extracted from the first frame of the video to the second artificial neural network. Furthermore, the apparatus includes means for processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame. Furthermore, the apparatus includes means for updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video.

[0014]

[0014] In one aspect of the disclosure, a non-transitory computer-readable medium is provided. The computer-readable medium has encoded thereon program code for processing a video. The program code, executed by a processor, includes code for receiving a video as an input at a first layer of a first artificial neural network and a second artificial neural network. The first artificial neural network has fewer channels than the second artificial neural network. The program code also includes code for processing a first frame of the video through the first artificial neural network to generate a first label. The first artificial neural network feeds intermediate features extracted from the first frame of the video to the second artificial neural network. Furthermore, the program code includes code for processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame. Furthermore, the program code includes code for updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video.

[0015]

[0015] Additional features and advantages of the present disclosure are described below. Those skilled in the art will appreciate that the present disclosure may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present disclosure. Those skilled in the art will also appreciate that such equivalent constructions do not depart from the teachings of the present disclosure as set forth in the appended claims. The novel features believed to characterize the present disclosure, both as to its organization and its method of operation, together with further objects and advantages, will be better understood upon consideration of the following description in conjunction with the accompanying figures. It should be expressly understood, however, that each of the figures is provided for the purpose of illustration and description only, and is not intended to define the limits of the present disclosure.

[0016]

[0016] The features, nature, and advantages of the present disclosure will become more apparent from the detailed description of the embodiments set forth below, taken in conjunction with the drawings in which like reference characters refer to like elements throughout. [Brief description of the drawings]

[0017] [Figure 1]

[0017] FIG. 1 illustrates an exemplary implementation of a neural network using a system-on-chip (SoC) including a general-purpose processor in accordance with some aspects of the present disclosure. [Figure 2A]

[0018] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 2B] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 2C] FIG. 1 illustrates a neural network according to an aspect of the present disclosure. [Figure 2D] FIG. 1 illustrates an example deep convolutional neural network (DCN), in accordance with aspects of the present disclosure. [Diagram 3]

[0019] FIG. 1 is a block diagram illustrating an example deep convolutional network (DCN), in accordance with aspects of the present disclosure. [Figure 4]

[0020] FIG. 1 is a block diagram illustrating an example software architecture that may modularize artificial intelligence (AI) functions. [Diagram 5]

[0021] 1 is a block diagram illustrating an example architecture for processing video, according to aspects of the present disclosure. [Figure 6]

[0022] 1 is a block diagram illustrating an example architecture for processing video, according to aspects of the present disclosure. [Figure 7]

[0023] 1 is a block diagram illustrating an example architecture for processing video, according to aspects of the present disclosure. [Figure 8]

[0024] 1 is a more detailed diagram illustrating an example architecture for processing video, according to aspects of the present disclosure. [Figure 9]

[0025] 1 is a flow diagram illustrating a method for processing video according to an aspect of the present disclosure. [Figure 10]

[0026] 1 is a flow diagram illustrating a method for processing video according to an aspect of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018]

[0027] The detailed description of the embodiments of the invention described below with reference to the accompanying drawings illustrates various configurations and does not represent the only configurations in which the described concepts may be practiced. The detailed description of the embodiments of the invention includes specific details for the purpose of providing a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. In some instances, well-known structures and components are shown in block diagram form in order to avoid obscuring such concepts.

[0019]

[0028] Based on these teachings, one skilled in the art should appreciate that the scope of the present disclosure covers any aspect of the present disclosure, whether implemented independently of or in combination with other aspects of the present disclosure. For example, an apparatus may be implemented or a method may be practiced using any number of the described aspects. Furthermore, the scope of the present disclosure is intended to cover such an apparatus or method implemented using other structures, functions, or structures and functions in addition to or other than the various aspects of the present disclosure described. It should be understood that any aspect of the present disclosure disclosed may be embodied by one or more elements of a claim.

[0020]

[0029] The word "exemplary" is used to mean "serving as an example, instance, or illustration." Any aspect described as "exemplary" is not necessarily to be construed as preferred or advantageous over other aspects.

[0021]

[0030] Although specific aspects are described, many variations and permutations of these aspects fall within the scope of the present disclosure. Although some benefits and advantages of the preferred aspects are described, the scope of the present disclosure is not limited to any particular benefit, use, or purpose. Rather, the aspects of the present disclosure are intended to be broadly applicable to a variety of technologies, system configurations, networks, and protocols, some of which are illustrated by way of example in the figures and the following description of the preferred aspects. The detailed description and drawings are merely illustrative rather than limiting of the present disclosure, the scope of which is defined by the appended claims and their equivalents.

[0022]

[0031] Neural networks can be used to solve complex problems, but the time it takes for the network to complete a task can be long, as the network size and the number of calculations that can be performed to produce a solution can be large. Furthermore, the computational cost of deep neural networks can be an issue, as these tasks may be performed on mobile devices that may have limited computing power.

[0023]

[0032] Conventional deep learning-based video processing methods or models may produce inconsistent output over time. In some cases, the inconsistency may be observed in the form of flickering or other inconsistencies in the display. Temporally inconsistent output (e.g., flickering) may degrade user experience and enjoyment, as well as system stability.

[0024]

[0033] One reason for temporally inconsistent output (e.g., predictions) is that the neural network may provide uncertain predictions, for example, when the output is around 0.5. In such cases, the neural network may produce predictions in an arbitrary manner. Thus, the segmentation accuracy may be significantly reduced for some frames in the video stream (between more accurately segmented frames). Thus, similar-looking image regions processed by the neural network may yield different predictions. Moreover, the reduction in segmentation accuracy may adversely affect performance.

[0025]

[0034] To address this issue, aspects of the present disclosure are directed to online (e.g., at test time) adaptation of segmentation models. A segmentation network for processing a video may be updated while processing the video. In some aspects, an auxiliary network may be incorporated along with the segmentation network to perform network updates while the segmentation network continues to process the video.

[0026]

[0035] 1 illustrates an exemplary implementation of a system-on-chip (SoC) 100 that may include a central processing unit (CPU) 102 or a multi-core CPU configured for video processing using an artificial neural network (e.g., a neural end-to-end network). Variables (e.g., neural signals and synaptic weights), system parameters associated with the computing device (e.g., a neural network with weights), delays, frequency bin information, and task information may be stored in a memory block associated with a neural processing unit (NPU) 108, in a memory block associated with the CPU 102, in a memory block associated with a graphics processing unit (GPU) 104, in a memory block associated with a digital signal processor (DSP) 106, in a memory block 118, or distributed across multiple blocks. Instructions executed in the CPU 102 may be loaded from a program memory associated with the CPU 102 or from the memory block 118.

[0027]

[0036] The SoC 100 may also include additional processing blocks adapted for specific functions, such as the GPU 104, the DSP 106, a connectivity block 110 that may include fifth generation (5G) connectivity, fourth generation long term evolution (4G LTE) connectivity, Wi-Fi connectivity, USB connectivity, Bluetooth connectivity, etc., and a multimedia processor 112 that may, for example, detect and recognize gestures. In one implementation, the NPU 108 is implemented in the CPU 102, the DSP 106, and / or the GPU 104. The SoC 100 may also include a sensor processor 114, an image signal processor (ISP) 116, and / or a navigation module 120 that may include a global positioning system.

[0028]

[0037] The SoC 100 may be based on the ARM instruction set. In one aspect of the disclosure, the instructions loaded into the general purpose processor 102 may include code for receiving a video as an input in a first layer of an artificial neural network (ANN). The general purpose processor 102 may also include code for processing a first frame of the video to generate a first label. The general purpose processor 102 may further include code for updating the artificial neural network based on the first label. The update is performed while concurrently processing a second frame of the video.

[0029]

[0038] In one aspect of the disclosure, the instructions loaded into the general purpose processor 102 may include code for receiving a video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network. The first artificial neural network has fewer channels than the second artificial neural network. The general purpose processor 102 may also include code for processing a first frame of the video through the first artificial neural network to generate a first label. The first artificial neural network feeds intermediate features extracted from the first frame of the video to the second artificial neural network. The general purpose processor 102 may further include code for processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame. The general purpose processor 102 may further include code for updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video.

[0030]

[0039] Deep learning architectures may perform object recognition tasks by learning to represent inputs at successively higher levels of abstraction at each layer, thereby accumulating useful feature representations of the input data. In this way, deep learning addresses a major bottleneck of traditional machine learning. Prior to the advent of deep learning, machine learning approaches to object recognition problems may have relied heavily on human-designed features, possibly in combination with shallow classifiers. A shallow classifier may be, for example, a two-class linear classifier in which a weighted sum of feature vector components may be compared to a threshold to predict which class an input belongs to. Human-designed features may be templates or kernels adapted to a particular problem domain by an engineer with domain expertise. In contrast, deep learning architectures learn to represent features that are similar to those that a human engineer might design, but may do so through training. Furthermore, deep networks may learn to represent and recognize new types of features that humans may not have considered.

[0031]

[0040] Deep learning architectures may learn a hierarchy of features. For example, when presented with visual data, a first layer may learn to recognize relatively simple features in the input stream, such as edges. In another example, when presented with auditory data, the first layer may learn to recognize spectral power at specific frequencies. A second layer, taking the output of the first layer as input, may learn to recognize combinations of features, such as simple shapes in the case of visual data, or combinations of sounds in the case of auditory data. For example, higher layers may learn to represent complex shapes in visual data, or words in auditory data. Even higher layers may learn to recognize common visual objects or spoken phrases.

[0032]

[0041] Deep learning architectures can work particularly well when applied to problems that have a natural hierarchical structure. For example, motor vehicle classification can benefit from first learning to recognize wheels, windshields, and other features. These features can be combined in different ways in higher layers to recognize cars, trucks, and planes.

[0033]

[0042] Neural networks can be designed with a variety of connectivity patterns. In feedforward networks, information is passed from lower layers to higher layers, with each neuron in a given layer communicating to a neuron in a higher layer. As explained above, in successive layers of a feedforward network, a hierarchical representation can be accumulated. Neural networks can also have recurrent or (also called top-down) feedback connections. In recurrent connections, the output from a neuron in a given layer can be communicated to another neuron in the same layer. Recurrent architectures can be useful for recognizing patterns across two or more of the input data chunks that are delivered sequentially to the neural network. Connections from neurons in a given layer to neurons in a lower layer are called feedback (or top-down) connections. Networks with many feedback connections can be useful when recognition of high-level concepts can help discriminate certain low-level features of the input.

[0034]

[0043] The connections between layers of a neural network can be fully connected or locally connected. FIG. 2A shows an example of a fully connected neural network 202. In the fully connected neural network 202, a neuron in a first layer can communicate its output to every neuron in a second layer, such that each neuron in the second layer receives input from every neuron in the first layer. FIG. 2B shows an example of a locally connected neural network 204. In the locally connected neural network 204, a neuron in a first layer can be connected to a limited number of neurons in the second layer. More generally, the locally connected layer of the locally connected neural network 204 can be configured such that each neuron in the layer has the same or similar connectivity pattern, but with connection strengths that can have different values ​​(e.g., 210, 212, 214, and 216). The connectivity pattern of the locally connected can result in spatially separate receptive fields in the upper layers, since the upper layer neurons in a given region can receive inputs that are conditioned through training to the properties of a limited portion of the total inputs to the network.

[0035]

[0044] One example of a locally connected neural network is a convolutional neural network. Figure 2C shows an example of a convolutional neural network 206. The convolutional neural network 206 may be configured such that the connection strengths associated with the inputs for each neuron in the second layer are shared (e.g., 208). Convolutional neural networks may be suitable for problems where the spatial location of the inputs is meaningful.

[0036]

[0045] One type of convolutional neural network is the deep convolutional network (DCN). Figure 2D shows a detailed example of a DCN 200 designed to recognize visual features from images 226 input from an image capture device 230, such as an on-board camera. The DCN 200 of this example may be trained to identify traffic signs and numbers given on traffic signs. Of course, the DCN 200 may be trained for other tasks, such as identifying lane markings or identifying traffic signals.

[0037]

[0046] The DCN 200 may be trained using supervised learning. During training, the DCN 200 may be presented with an image, such as an image 226 of a speed limit sign, and then a forward pass may be calculated to produce the output 222. The DCN 200 may include a feature extraction section and a classification section. Upon receiving the image 226, the convolution layer 232 may apply a convolution kernel (not shown) to the image 226 to generate the first set of feature maps 218. As an example, the convolution kernel for the convolution layer 232 may be a 5×5 kernel that generates a 28×28 feature map. In this example, four different feature maps are generated in the first set of feature maps 218, so four different convolution kernels were applied to the image 226 in the convolution layer 232. A convolution kernel may also be referred to as a filter or a convolution filter.

[0038]

[0047] The first set of feature maps 218 may be subsampled by a max pooling layer (not shown) to generate a second set of feature maps 220. The max pooling layer reduces the size of the first set of feature maps 218. That is, the size of the second set of feature maps 220, such as 14×14, is smaller than the size of the first set of feature maps 218, such as 28×28. The reduced size provides similar information to subsequent layers while reducing memory consumption. The second set of feature maps 220 may be further convolved via one or more subsequent convolutional layers (not shown) to generate one or more subsequent sets of feature maps (not shown).

[0039]

[0048] In the example of FIG. 2D , the second set of feature maps 220 are convolved to generate a first feature vector 224. Moreover, the first feature vector 224 is further convolved to generate a second feature vector 228. Each feature in the second feature vector 228 may include a number corresponding to a possible feature of the image 226, such as “sign”, “60”, and “100”. A softmax function (not shown) may convert the numbers in the second feature vector 228 into probabilities. Thus, the output 222 of the DCN 200 is the probability that the image 226 contains one or more features.

[0040]

[0049] In this example, the probability in output 222 for "sign" and "60" is higher than the probability of others of output 222, such as "30", "40", "50", "70", "80", "90", and "100". Before training, the output 222 produced by DCN 200 may be inaccurate. Thus, an error may be calculated between output 222 and a target output. The target output is the ground truth of image 226 (e.g., "sign" and "60"). The weights of DCN 200 may then be adjusted so that the output 222 of DCN 200 is more closely aligned with the target output.

[0041]

[0050] To adjust the weights, the learning algorithm may calculate a gradient vector for the weights. The gradient may indicate the amount by which the error increases or decreases if the weights are adjusted. In the top layer, the gradient may correspond directly to the values ​​of the weights connecting the activated neurons in the penultimate layer and the neurons in the output layer. In lower layers, the gradient may depend on the values ​​of the weights and the calculated error gradient of the upper layer. The weights may then be adjusted to reduce the error. This manner of adjusting weights is sometimes called "back propagation" because it involves a "backward pass" through the neural network.

[0042]

[0051] In practice, the error gradient of the weights may be computed over a small number of examples such that the computed gradient approximates the true error gradient. This approximation method is sometimes called stochastic gradient descent. Stochastic gradient descent may be repeated until the achievable error rate of the overall system no longer decreases or until the error rate reaches a target level. After training, the DCN may be presented with new images and a forward pass through the network may result in an output 222, which may be considered the inference or prediction of the DCN.

[0043]

[0052] A deep belief network (DBN) is a probabilistic model with multiple layers of hidden nodes. DBNs can be used to extract hierarchical representations of a training dataset. DBNs can be obtained by stacking layers of Restricted Boltzmann Machines (RBMs). RBMs are a type of artificial neural network that can learn probability distributions over a set of inputs. RBMs are often used in unsupervised learning because they can learn probability distributions in the absence of information about the class into which each input should be categorized. Using a hybrid unsupervised and supervised paradigm, the lower RBM of the DBN can be trained in an unsupervised manner and act as a feature extractor, and the upper RBM can be trained in a supervised manner (on the joint distribution of inputs from the previous layer and the target class) and act as a classifier.

[0044]

[0053] Deep Convolutional Networks (DCNs) are networks of convolutional networks, composed of additional pooling and normalization layers. DCNs have achieved state-of-the-art performance for many tasks. DCNs can be trained using supervised learning, where both the input and output targets are known for many examples and are used to modify the network weights by using gradient descent methods.

[0045]

[0054] The DCN may be a feedforward network. Furthermore, as described above, the connections from neurons in the first layer of the DCN to groups of neurons in the next higher layer are shared across the neurons in the first layer. The feedforward and shared connections of the DCN may be exploited for high speed processing. The computational burden of the DCN may be much less than that of a similarly sized neural network with, for example, recurrent or feedback connections.

[0046]

[0055] The processing of each layer of the convolutional network may be considered as a spatially invariant template or basis projection. If the input is initially decomposed into multiple channels, such as the red, green, and blue channels of a color image, the convolutional network trained on that input may be considered to be three-dimensional, with two spatial dimensions along the axes of the image and a third dimension capturing color information. The output of the convolutional combinations may be considered to form a feature map in the subsequent layer, where each element of the feature map (e.g., 220) may receive inputs from various neurons in the previous layer (e.g., feature map 218) and from each of the multiple channels. The values ​​in the feature map may be further processed using nonlinearities, such as rectification, max(0,x), etc. Values ​​from neighboring neurons may be further pooled, which corresponds to downsampling and may provide further local invariance and dimensionality reduction. Normalization, which corresponds to whitening, may also be applied by lateral inhibition between neurons in the feature map.

[0047]

[0056] The performance of deep learning architectures can improve as more labeled data points become available or as computational power increases. Modern deep neural networks are routinely trained with computing resources thousands of times greater than those available to the average researcher only 15 years ago. New architectures and training paradigms can further improve deep learning performance. Rectified linear units can reduce a training problem known as vanishing gradients. New training techniques can reduce over-fitting and thus enable larger models to achieve better generalization. Encapsulation techniques can extract data in a given receptive field, further improving overall performance.

[0048]

[0057] 3 is a block diagram illustrating a deep convolutional network 350. The deep convolutional network 350 may include multiple different types of layers based on connectivity and weight sharing. As shown in FIG. 3, the deep convolutional network 350 includes convolutional blocks 354A, 354B. Each of the convolutional blocks 354A, 354B may be composed of a convolutional layer (CONV) 356, a normalization layer (LNorm) 358, and a max pooling layer (MAX POOL) 360.

[0049]

[0058] The convolution layer 356 may include one or more convolution filters, which may be applied to the input data to generate feature maps. Although only two of the convolution blocks 354A, 354B are shown, the disclosure is not so limited and instead any number of convolution blocks 354A, 354B may be included in the deep convolution network 350 according to design preferences. The normalization layer 358 may normalize the output of the convolution filters. For example, the normalization layer 358 may perform whitening or lateral suppression. The max pooling layer 360 may perform downsampling aggregation across space for local invariance and dimensionality reduction.

[0050]

[0059] For example, the parallel filter bank of the deep convolutional network may be loaded into the CPU 102 or GPU 104 of the SoC 100 to achieve high performance and low power consumption. In alternative embodiments, the parallel filter bank may be loaded into the DSP 106 or ISP 116 of the SoC 100. Additionally, the deep convolutional network 350 may access other processing blocks that may be present on the SoC 100, such as the sensor processor 114 and the navigation module 120, which are dedicated to sensors and navigation, respectively.

[0051]

[0060] The deep convolutional network 350 may also include one or more fully connected layers 362 (FC1 and FC2). The deep convolutional network 350 may further include a logistic regression (LR) layer 364. Between each layer 356, 358, 360, 362, 364 of the deep convolutional network 350, there are weights (not shown) to be updated. The output of each of the layers (e.g., 356, 358, 360, 362, 364) may serve as an input of a subsequent layer of the layers (e.g., 356, 358, 360, 362, 364) in the deep convolutional network 350 to learn a hierarchical feature representation from the input data 352 (e.g., image, audio, video, sensor data, and / or other input data) provided in the first of the convolutional blocks 354A. The output of the deep convolutional network 350 is a classification score 366 for the input data 352. The classification score 366 may be a set of probabilities, where each probability is the probability that the input data contains a feature from the set of features.

[0052]

[0061] 4 is a block diagram illustrating an example software architecture 400 that may modularize artificial intelligence (AI) functions. Using that architecture, applications may be designed that may cause various processing blocks (e.g., CPU 422, DSP 424, GPU 426, and / or NPU 428) of a system on chip (SoC) 420 to support adaptive rounding as disclosed for post-training quantization for AI applications 402 according to aspects of the present disclosure.

[0053]

[0062] The AI ​​application 402 may be configured to invoke functions defined in user space 404, which may provide, for example, scene detection and recognition indicative of the location in which the device is currently operating. The AI ​​application 402 may configure the microphone and camera differently depending on whether the recognized scene is, for example, an office, an auditorium, a restaurant, or an outdoor environment such as a lake. The AI ​​application 402 may make requests to compiled program code associated with libraries defined in the AI ​​function application programming interface (API) 406. The requests may ultimately rely on the output of a deep neural network configured to provide an inference response based on, for example, the video and positioning data.

[0054]

[0063] Additionally, a runtime engine 408, which may be compiled code of a runtime framework, may be accessible to the AI ​​application 402. The AI ​​application 402 may, for example, cause the runtime engine to request inference at specific time intervals or triggered by an event detected by the application's user interface. When caused to provide an inference response, the runtime engine may in turn signal an operating system in the operating system (OS) space, such as kernel 412, running on the SoC 420. The operating system may in turn cause successive relaxations of quantization to be performed on the CPU 422, DSP 424, GPU 426, NPU 428, or some combination thereof. The CPU 422 may be accessed directly by the operating system, and other processing blocks may be accessed through drivers, such as drivers 414, 416, or 418 for the DSP 424, GPU 426, or NPU 428, respectively. In an illustrative example, the deep neural network may be configured to run on a combination of processing blocks, such as a CPU 422, a DSP 424, and a GPU 426, or may be run on an NPU 428.

[0055]

[0064] An application 402 (e.g., an AI application) may be configured to invoke functions defined in user space 404, which may provide, for example, scene detection and recognition indicative of the location in which the device is currently operating. The application 402 may configure the microphone and camera differently depending on whether the recognized scene is, for example, an office, an auditorium, a restaurant, or an outdoor environment such as a lake. The application 402 may make a request to compiled program code associated with a library defined in SceneDetect application programming interface (API) 406 to provide an estimate of the current scene. The request may ultimately rely on the output of, for example, a differential neural network configured to provide a scene estimate based on video and positioning data.

[0056]

[0065] Additionally, a runtime engine 408, which may be compiled code of a runtime framework, may be accessible to the application 402. The application 402 may, for example, have the runtime engine request scene estimation at specific time intervals or triggered by an event detected by the application's user interface. When the scene has been estimated, the runtime engine may in turn signal an operating system 410, such as a kernel 412, running on the SoC 420. The operating system 410 may in turn cause the computations to be performed on the CPU 422, the DSP 424, the GPU 426, the NPU 428, or some combination thereof. The CPU 422 may be accessed directly by the operating system, and other processing blocks may be accessed through drivers, such as drivers 414-418 for the DSP 424, for the GPU 426, or for the NPU 428. In an illustrative example, the differential neural network may be configured to run on a combination of processing blocks, such as the CPU 422 and the GPU 426, or may run on the NPU 428.

[0057]

[0066] Aspects of the present disclosure are directed to deep neural network model transplantation using adversarial function approximation.

[0058]

[0067] FIG. 5 is a block diagram illustrating an example architecture 500 for processing a video according to an aspect of the present disclosure. As shown in FIG. 5, a segmentation network 502 receives a video 504 as an input. The segmentation network 502 may be, for example, a convolutional neural network, such as the deep convolutional network 350 shown in FIG. 3. The video 504 is divided in frames (e.g., raw t1, raw t2, raw t3, raw t4, raw t5, and raw t6). The segmentation network 502 processes each frame of the video 504 successively and generates an output 506 in the form of a segment (e.g., seg t1, seg t2, seg t3, seg t4, seg t5, and seg t6). While processing the video 504, the segmentation network 502 may also be updated. As shown in FIG. 5, after generating each output segment, the segmentation network 502 is updated via a back path (eg, 508a-508f).

[0059]

[0068] FIG. 6 is a block diagram illustrating an example architecture 600 for processing a video according to an aspect of the present disclosure. As shown in FIG. 6, a segmentation network 604 receives a video 602 as an input. The segmentation network 604 may be configured to operate similarly to the segmentation network 502 of FIG. 5. The segmentation network 604 processes each frame of the video 602 and generates an output segment 606. Each output segment 606 is subjected to an argmax operation to generate a pseudo label 608 and an output label 612. Applying cross-entropy (CE) as a loss function, a negative log-likelihood may be calculated based on the pseudo label 608 and the output label 612. In turn, the segmentation network 604 may be updated via backpropagation (BP) to reduce the CE loss. In some aspects, the argmax may be used to determine the output label 612, and may be complemented by a confidence value calculated using, for example, a softmax likelihood or a logit function. Moreover, in some aspects, the confidence value may be a best estimate or a sub-best estimate label ratio.

[0060]

[0069] FIG. 7 is a block diagram illustrating an example architecture 700 for processing video according to an aspect of the disclosure. With reference to FIG. 7, the example architecture 700 includes a segmentation network 704 and an auxiliary network 706. The auxiliary network 706 may have an overall architecture similar to that of the segmentation network 704. However, the auxiliary network 706 is smaller than the segmentation network 704 (e.g., 1 / 10 the size). For example, in some aspects, the auxiliary network 706 may be configured with fewer channels than the segmentation network 704 (e.g., the auxiliary network 706 may have 18 channels and the segmentation network 704 may have 48 channels) or may operate at a lower resolution than the segmentation network 704. In one example, the auxiliary network 706 may run on a 32-bit CPU (e.g., CPU 102 in FIG. 1) or GPU (e.g., GPU 104 in FIG. 1), and the segmentation network 704 may run on an 8-bit DSP (e.g., DSP 106 in FIG. 1).

[0061]

[0070] In operation, the segmentation network 704 and the auxiliary network 706 each receive as an input a video 702. The segmentation network 704 processes each frame of the video 702 and produces output segments 710. The auxiliary network 706 processes each frame of the video 702 and produces segments 708.

[0062]

[0071] Additionally, the auxiliary network 706 feeds the intermediate features to the segmentation network 704. The segmentation network 704 processes the intermediate features generated in the layers of the segmentation network and the aggregated intermediate features (as indicated by the "+" sign) to compute output segments 710. Each output segment 710 is subjected to an argmax operation to generate a pseudo label 712. Similarly, each segment 708 is subjected to an argmax operation to generate an output label 714.

[0063]

[0072] The pseudo labels 712 and the output labels 714 may be used to compute a mutual entropy loss and to update, e.g., via backpropagation, the auxiliary network 706 while the segmentation network 704 continues to process the video 702 in the forward pass. Because the updates are performed on the smaller auxiliary network 706 rather than the segmentation network 704, the computational cost may be significantly reduced relative to that of the segmentation network 604 of FIG.

[0064]

[0073] FIG. 8 is a more detailed diagram 800 illustrating the example architecture 700 of FIG. 7 according to an aspect of the disclosure. Referring to FIG. 8, a segmentation network 804 and an auxiliary network 806 are shown. In the example of FIG. 8, the segmentation network 804 and the auxiliary network 806 may have similar architectures. For example, the segmentation network 804 and the auxiliary network 806 may be configured as autoencoders. As described above with respect to FIG. 7, the auxiliary network 806 may be smaller than the segmentation network 804 (e.g., 1 / 10 the size). For example, in some aspects, the auxiliary network 806 may be configured with fewer channels (e.g., the auxiliary network 806 may have 18 channels and the segmentation network 804 may have 48 channels). Furthermore, in some aspects, the auxiliary network 806 may operate at a lower resolution than the segmentation network 804. 8, the auxiliary network 806 and the segmentation network 804 both receive a video 802 as an input. However, while the segmentation network 804 downsamples each frame of the video 802 three times, the auxiliary network 806 downsamples each frame of the video 802 four times to produce a lower resolution of each frame of the video compared to the segmentation network 804.

[0065]

[0074] Upon downsampling the frames of the video 802, the segmentation network 804 and the auxiliary network 806 each pass their respective lower resolution frames through successive layers of convolutional filters (e.g., convolutional coding blocks) to extract features. Further downsampling may be performed via transition blocks (e.g., Transition 1, Transition 2, and Transition 3) of the segmentation network 804 and the auxiliary network 806, respectively, to extract lower resolution features.

[0066]

[0075] Additionally, the auxiliary network 806 may provide intermediate features (e.g., 808, 810) to the segmentation network 804. The intermediate features (e.g., 808, 810) may be mapped, combined, or aggregated with the intermediate features of the segmentation network 804 (to account for a larger number of channels in the segmentation network 804). The segmentation network 804 and the auxiliary network 806 each produce a set of output features (e.g., 812, 814), which may be upsampled, concatenated, and provided to the decoders (816, 818).

[0067]

[0076] Decoders 816 and 818, in turn, process the output features (e.g., 812, 814) to generate output segments 820 and 822. Each of the output segments 820, 822 is subjected to an argmax operation to generate a label 824.

[0068]

[0077] Furthermore, the auxiliary network 806 may be updated using backpropagation. In particular, the auxiliary network 806 is updated online (e.g., at test time) as the segmentation network 804 continues to process the video 802. In doing so, temporal inconsistency may be reduced (e.g., an intersection over union metric may be increased). Thus, the flickering effect observed in the video 802 may be reduced. Furthermore, computational costs and processing times may be reduced.

[0069]

[0078] 9 illustrates a method 900 for processing a video according to an aspect of the present disclosure. As shown in FIG. 9, at block 902, the method 900 receives a video input at a first layer of an artificial neural network (ANN). The first artificial neural network may be a convolutional neural network, such as the deep convolutional network 350 of FIG. 3. As described with reference to FIG. 5, a segmentation network 502 receives a video 504 as an input.

[0070]

[0079] At block 904, the method 900 processes a first frame of the video to generate a first label. As described with reference to FIG. 5, the segmentation network 502 sequentially processes each frame of the video 504 and generates an output 506 in the form of a segment (e.g., seg t1).

[0071]

[0080] At block 906, the method 900 updates the artificial neural network based on the first label, where the updating is performed while concurrently processing a second frame of the video. For example, as shown in FIG. 5, after generating each output segment, the segmentation network is updated via a back path (e.g., 508a-508f).

[0072]

[0081] FIG. 10 illustrates a method 1000 for processing a video according to an embodiment of the present disclosure. As illustrated in FIG. 10, in block 1002, the method 1000 receives a video as an input in a first layer of a first artificial neural network (ANN) and a second artificial neural network. The first artificial neural network has fewer channels than the second artificial neural network. For example, as illustrated in FIG. 7, an exemplary architecture 700 includes a segmentation network 704 and an auxiliary network 706. The segmentation network 704 and the auxiliary network 706 each receive a video 702 as an input. Furthermore, the auxiliary network 706 may have an architecture similar to that of the segmentation network 704. For example, in some aspects, the auxiliary network 706 may be configured with fewer channels than the segmentation network 704 (e.g., the auxiliary network 706 may have 18 channels and the segmentation network 704 may have 48 channels) or may operate at a lower resolution than the segmentation network 704.

[0073]

[0082] At block 1004, the method 1000 processes the first frame of the video through a first artificial neural network to generate a first label. The first artificial neural network supplies intermediate features extracted from the first frame of the video to a second artificial neural network. For example, as described with reference to FIG. 7, the auxiliary network 706 processes each frame of the video and generates segments 708. The auxiliary network 706 supplies the intermediate features to the segmentation network 704. The segmentation network 704 processes the intermediate features, which are aggregated to compute output segments 710.

[0074]

[0083] At block 1006, the method 1000 processes the first frame of the video through a second artificial neural network to generate a second label based on the intermediate features and the first frame. For example, the segmentation network 704 processes each frame of the video 702 and generates output segments 710.

[0075]

[0084] At block 1008, the method 1000 updates the first artificial neural network based on the first label while the second artificial neural network concurrently processes the second frame of the video. As described with reference to Figure 7, the pseudo label 712 may be used to update the auxiliary network 706, e.g., via backpropagation, while the segmentation network 704 continues to process the video 702 in the forward pass.

[0076]

[0085] In one aspect, the receiving means, processing means, and / or updating means may be the CPU 102, the GPU 104, the DSP 106 program memory associated with the CPU 102, the dedicated memory block 118, the fully connected layer 362, the NPU 428, and / or the routing connection processing unit 216 configured to perform the recited functions. In another configuration, the above-mentioned means may be any module or any device configured to perform the recited functions by the above-mentioned means.

[0077]

[0086] The various operations of the methods described above may be performed by any suitable means capable of performing the corresponding functions. Those means may include various (one or more) hardware and / or software components and / or modules, including but not limited to circuits, application specific integrated circuits (ASICs), or processors. Generally, where there are operations illustrated in the figures, those operations may have corresponding counterpart means-plus-function components with similar numbering.

[0078]

[0087] Implementation examples are provided in the following numbered clauses. 1. A method for processing a video, comprising: receiving a video as an input at a first layer of an artificial neural network (ANN); processing a first frame of the video to generate a first label; updating the artificial neural network based on the first label, and the updating is performed while concurrently processing a second frame of the video. A method comprising: 2. The method of claim 1, further comprising applying the first label in a backward pass of the artificial neural network to update the artificial neural network. 3. The method of any of clauses 1 to 2, further comprising generating a second label based on the second frame, and concurrent processing is performed to reduce temporal inconsistency between the first label and the second label. 4. A method for processing a video, comprising: receiving video as an input at a first artificial neural network (ANN) and a first layer of a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; processing a first frame of the video through a first artificial neural network to generate a first label; and the first artificial neural network feeding intermediate features extracted from the first frame of the video to a second artificial neural network. processing a first frame of the video through a second artificial neural network to generate a second label based on the intermediate features and the first frame; updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes a second frame of the video; A method comprising: 5. The method of claim 4, further comprising generating a third label via a second artificial neural network based on the second frame, wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced. 6. The method of any of clauses 4 to 5, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network. 7. An apparatus for processing video, comprising: Memory, at least one processor coupled to a memory; At least one processor comprises: receiving a video as an input at a first layer of an artificial neural network (ANN); processing a first frame of the video to generate a first label; updating the artificial neural network based on the first label, wherein the updating is performed while concurrently processing a second frame of the video. An apparatus configured to: 8. The apparatus of clause 7, wherein the at least one processor is further configured to apply the first label in a backward pass of the artificial neural network to update the artificial neural network. 9. The apparatus of any of clauses 7 to 8, wherein at least one processor is further configured to generate a second label based on the second frame, and wherein concurrent processing is performed to reduce temporal inconsistency between the first label and the second label. 10. An apparatus for processing video, comprising: Memory, at least one processor coupled to a memory; At least one processor comprises: receiving video as an input at a first artificial neural network (ANN) and a first layer of a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; processing a first frame of the video through a first artificial neural network to generate a first label; and the first artificial neural network feeding intermediate features extracted from the first frame of the video to a second artificial neural network. processing a first frame of the video through a second artificial neural network to generate a second label based on the intermediate features and the first frame; updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes a second frame of the video; An apparatus configured to: 11. The apparatus of clause 10, wherein the at least one processor is further configured to generate a third label via a second artificial neural network based on the second frame, and the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced. 12. The apparatus of claim 10, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network. 13. An apparatus for processing video, comprising: means for receiving a video as an input at a first layer of an artificial neural network (ANN); means for processing a first frame of the video to generate a first label; means for updating the artificial neural network based on the first label, wherein the updating is performed while concurrently processing a second frame of the video. An apparatus comprising: 14. The apparatus of claim 13, further comprising means for applying the first label in a backward pass of the artificial neural network to update the artificial neural network. 15. The apparatus of any of clauses 13 to 14, further comprising means for generating a second label based on the second frame, and wherein concurrent processing is performed to reduce temporal inconsistency between the first label and the second label. 16. An apparatus for processing video, comprising: means for receiving video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; means for processing the first frame of the video through a first artificial neural network to generate a first label, the first artificial neural network feeding intermediate features extracted from the first frame of the video to a second artificial neural network; means for processing a first frame of the video through a second artificial neural network to generate a second label based on the intermediate features and the first frame; means for updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes a second frame of the video; An apparatus comprising: 17. The apparatus of clause 16, further comprising means for generating a third label via a second artificial neural network based on the second frame, wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced. 18. Apparatus according to any of clauses 16 to 17, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network. 19. A non-transitory computer-readable medium having encoded thereon program code for processing video, the program code being executed by a processor; program code for receiving a video as an input at a first layer of an artificial neural network (ANN); program code for processing a first frame of a video to generate a first label; program code for updating the artificial neural network based on the first label, the updating being performed while concurrently processing a second frame of the video; 1. A non-transitory computer-readable medium comprising: 20. The non-transitory computer-readable medium of claim 19, further comprising program code for applying the first label in a backward pass of the artificial neural network to update the artificial neural network. 21. The non-transitory computer-readable medium of any of clauses 19 to 20, further comprising program code for generating a second label based on the second frame, and wherein concurrent processing is performed to reduce temporal inconsistency between the first label and the second label. 22. A non-transitory computer-readable medium having encoded thereon program code for processing video, the program code being executed by a processor; program code for receiving video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; and program code for processing a first frame of the video through a first artificial neural network to generate a first label, the first artificial neural network feeding intermediate features extracted from the first frame of the video to a second artificial neural network. program code for processing a first frame of the video through a second artificial neural network to generate a second label based on the intermediate features and the first frame; program code for updating the first artificial neural network based on the first label while the second artificial neural network concurrently processes a second frame of the video; 1. A non-transitory computer-readable medium comprising: 23. The non-transitory computer-readable medium device of clause 22, further comprising program code for generating a third label via a second artificial neural network based on the second frame, and wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced. 24. The non-transitory computer-readable medium of any of clauses 22-23, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network.

[0079]

[0088] The term "determining" as used herein encompasses a wide variety of actions. For example, "determining" may include calculating, computing, processing, deriving, investigating, looking up (e.g., looking up in a table, database, or another data structure), ascertaining, and the like. Additionally, "determining" may include receiving (e.g., receiving information), accessing (e.g., accessing data in a memory), and the like. Additionally, "determining" may include resolving, selecting, choosing, establishing, and the like.

[0080]

[0089] As used herein, a phrase referring to "at least one of" a list of items refers to any combination of those items, including single members. As an example, "at least one of a, b, or c" is intended to encompass a, b, c, ab, ac, bc, and abc.

[0081]

[0090] The various example logic blocks, modules and circuits described in connection with this disclosure may be implemented or performed using a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array signal (FPGA) or other programmable logic device (PLD), discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the described functions. A general purpose processor may be a microprocessor, but alternatively, the processor may be any commercially available processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0082]

[0091] The steps of a method or algorithm described in connection with the present disclosure may be implemented directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in any form of storage medium known in the art. Some examples of storage media that may be used include random access memory (RAM), read only memory (ROM), flash memory, erasable programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM), registers, hard disks, removable disks, CD-ROMs, etc. A software module may comprise a single instruction, or many instructions, and may be distributed over several different code segments, among different programs, and across multiple storage media. A storage medium may be coupled to a processor such that the processor can read information from and write information to the storage medium. Alternatively, the storage medium may be integral to the processor.

[0083]

[0092] The methods disclosed herein comprise one or more steps or actions for achieving the described method. The steps and / or actions of the methods may be interchanged with one another without departing from the scope of the claims. In other words, unless a specific order of steps or actions is specified, the order and / or use of specific steps and / or actions may be modified without departing from the scope of the claims.

[0084]

[0093] The described functions may be implemented in hardware, software, firmware, or any combination thereof. When implemented in hardware, an exemplary hardware configuration may comprise a processing system in a device. The processing system may be implemented using a bus architecture. The bus may include any number of interconnected buses and bridges depending on the particular application and overall design constraints of the processing system. The bus may link various circuits together, including processors, machine-readable media, and bus interfaces. The bus interface may be used to connect a network adapter, among other things, to the processing system via the bus. The network adapter may be used to implement signal processing functions. In some aspects, a user interface (e.g., keypad, display, mouse, joystick, etc.) may also be connected to the bus. The bus may also link various other circuits, such as timing sources, peripherals, voltage regulators, power management circuits, etc., which are well known in the art and therefore will not be described further.

[0085]

[0094] The processor may be responsible for managing the bus and general processing, including the execution of software stored on the machine-readable medium. The processor may be implemented using one or more general-purpose and / or special-purpose processors. Examples include microprocessors, microcontrollers, DSP processors, and other circuitry capable of executing software. Software should be broadly construed to mean instructions, data, or any combination thereof, whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise. The machine-readable medium may include, by way of example, a random access memory (RAM), a flash memory, a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), registers, a magnetic disk, an optical disk, a hard drive, or any other suitable storage medium, or any combination thereof. The machine-readable medium may be embodied in a computer program product. The computer program product may comprise packaging materials.

[0086]

[0095] In a hardware implementation, the machine-readable medium may be part of a processing system separate from the processor. However, as one skilled in the art would readily appreciate, the machine-readable medium, or any portion thereof, may be external to the processing system. By way of example, the machine-readable medium may include a transmission line, a carrier wave modulated by data, and / or a computer product separate from the device, all of which may be accessed by the processor via a bus interface. Alternatively, or in addition, the machine-readable medium, or any portion thereof, may be integrated into the processor, such as a cache and / or a general-purpose register file. Although the various components described, such as local components, may be described as having specific locations, they may also be configured in various ways, such as some components configured as part of a distributed computing system.

[0087]

[0096] The processing system may be configured as a general-purpose processing system with one or more microprocessors providing processor functionality and external memory providing at least a portion of the machine-readable medium, all linked together with other support circuits via an external bus architecture. Alternatively, the processing system may include one or more neuromorphic processors for implementing the described neuron model and model of the neural system. As another alternative, the processing system may be implemented using an application specific integrated circuit (ASIC) with a processor, bus interface, user interface, support circuits, and at least a portion of the machine-readable medium integrated into a single chip, or using one or more field programmable gate arrays (FPGAs), programmable logic devices (PLDs), controllers, state machines, gate logic, discrete hardware components, or other suitable circuits, or any combination of circuits capable of performing the various functions described throughout this disclosure. Those skilled in the art will understand how the described functions of the processing system may be best implemented depending on the particular application and the overall design constraints imposed on the overall system.

[0088]

[0097] The machine-readable medium may comprise several software modules. The software modules include instructions that, when executed by the processor, cause the processing system to perform various functions. The software modules may include a transmitting module and a receiving module. Each software module may reside in a single storage device or be distributed across multiple storage devices. As an example, a software module may be loaded from a hard drive into RAM when a triggering event occurs. During execution of the software module, the processor may load some of the instructions into a cache to increase access speed. One or more cache lines may then be loaded into a general purpose register file for execution by the processor. When referring to the functionality of a software module below, it will be understood that such functionality is implemented by the processor when executing instructions from that software module. It should further be appreciated that aspects of the present disclosure may result in improvements in the functionality of a processor, computer, machine, or other system implementing such aspects.

[0089]

[0098] If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media, including any medium that enables transfer of a computer program from one place to another. A storage medium may be any available medium that can be accessed by a computer. By way of example, and not limitation, such computer-readable media may comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Furthermore, any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared (IR), radio, and microwave, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk and disc as used herein include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc, where disks typically reproduce data magnetically and discs reproduce data optically with lasers. Thus, in some aspects computer readable medium may comprise non-transitory computer readable medium (e.g., tangible media). Additionally, in other aspects computer readable medium may comprise transitory computer readable medium (e.g., signal). Combinations of the above should also be included within the scope of computer readable medium.

[0090]

[0099] Thus, some aspects may comprise a computer program product for performing the operations presented herein. For example, such a computer program product may comprise a computer-readable medium having stored thereon (and / or encoded) instructions executable by one or more processors to perform the described operations. In some aspects, the computer program product may include packaging materials.

[0091]

[0100] Furthermore, it should be appreciated that modules and / or other suitable means for implementing the described methods and techniques may be downloaded and / or otherwise obtained by a user terminal and / or base station, where applicable. For example, such devices may be coupled to a server to enable transfer of means for implementing the described methods. Alternatively, the various methods described may be provided by a storage means such that the user terminal and / or base station may obtain the various methods upon coupling or providing the storage means (e.g., RAM, ROM, physical storage medium such as a compact disk (CD) or floppy disk, etc.) to the device. Moreover, any other suitable technique for providing the described methods and techniques to a device may be utilized.

[0092]

[0101] It should be understood that the claims are not limited to the precise configuration and components illustrated above. Various modifications, changes and variations may be made in the arrangement, operation and details of the methods and apparatus described above without departing from the scope of the claims.

Claims

1. 1. A method for processing a video, comprising: receiving the video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; processing a first frame of the video through the first artificial neural network to generate a first label; and the first artificial neural network feeding intermediate features extracted from the first frame of the video to the second artificial neural network. processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame; updating the first artificial neural network based on the first label and the second label while the second artificial neural network concurrently processes a second frame of the video; A method comprising:

2. 2. The method of claim 1, further comprising generating a third label via the second artificial neural network based on the second frame, wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced.

3. 2. The method of claim 1, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network.

4. 1. An apparatus for processing video, comprising: Memory, at least one processor coupled to the memory; wherein the at least one processor: receiving the video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; processing a first frame of the video through the first artificial neural network to generate a first label; and the first artificial neural network feeding intermediate features extracted from the first frame of the video to the second artificial neural network. processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame; updating the first artificial neural network based on the first label and the second label while the second artificial neural network concurrently processes a second frame of the video; An apparatus configured to:

5. 5. The apparatus of claim 4, wherein the at least one processor is further configured to generate a third label via the second artificial neural network based on the second frame, and wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced.

6. 5. The apparatus of claim 4, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network.

7. A non-transitory computer readable medium having encoded thereon program code for processing video, the program code being executed by a processor; program code for receiving the video as an input at a first layer of a first artificial neural network (ANN) and a second artificial neural network, the first artificial neural network having fewer channels than the second artificial neural network; program code for processing a first frame of the video through the first artificial neural network to generate a first label; and the first artificial neural network feeding intermediate features extracted from the first frame of the video to the second artificial neural network. program code for processing the first frame of the video through the second artificial neural network to generate a second label based on the intermediate features and the first frame; program code for updating the first artificial neural network based on the first label and the second label while the second artificial neural network concurrently processes a second frame of the video; 1. A non-transitory computer-readable medium comprising:

8. 8. The non-transitory computer-readable medium device of claim 7, further comprising program code for generating a third label via the second artificial neural network based on the second frame, wherein the concurrent processing is performed such that temporal inconsistency between the second label and the third label is reduced.

9. 8. The non-transitory computer-readable medium of claim 7, wherein the first artificial neural network operates at a lower resolution than the second artificial neural network.

Images