Abstract: Computer implemented systems are described that implement a differentiable neural architecture search (DNAS) engine executing on one or more processors. The DNAS engine is configured with a stochastic super net defining a layer-wise search space having a plurality of candidate layers, each of the candidate layers specifying one or more operators for a neural network architecture. Further, the DNAS engine is configured to process training data to train weights for the operators in the stochastic super net based on a loss function representing a latency of the respective operator on a target platform, and to select a set of candidate neural network architectures from the trained stochastic super net. The DNAS engine may, for example, be configured to train the stochastic super net by traversing the layer-wise search space using gradient-based optimization of network architecture distribution.

Abstract: In one embodiment, a system accesses a probability model associated with an image depicting a body. The probability model includes probability values associated with regions of the image and each probability value represents a probability of the associated region of the image containing a particular body part. The system selects a subset (e.g., 3) of the probability values based on a comparison of the probability values. For each selected probability value, the system identifies surrounding probability values surrounding the selected probability value and computes a probabilistic maximum based on the selected probability value and the surrounding probability values. Each probabilistic maximum is associated with a location within the regions associated with the selected probability value and the surrounding probability values.

Abstract: In one embodiment, a system may access an image and generate a feature map for the image using a neural network. The system may identify regions of interest in the feature map. Regional feature maps may be generated for the regions of interest, respectively. Each of the regional feature maps has a first, a second, and a third dimension. The system may generate a first combined regional feature map by combining the regional feature maps. The combined regional feature map has a first, a second, and a third dimension. The system may generate a second combined regional feature map by processing the first combined regional feature map using one or more convolutional layers. The system may generate, for each of the regions of interest, information associated with an object instance based on a portion of the second combined regional feature map associated with that region of interest.

Abstract: In one embodiment, a system accesses pose probability models for predetermined parts of a body depicted in an image. Each of the pose probability models is configured for determining a probability of the associated predetermined body part being at a location in the image. The system determines a candidate pose that is defined by a set of coordinates representing candidate locations of the predetermined body parts. The system further determines a first probability score for the candidate pose based on the pose probability models and the set of coordinates of the candidate pose. A pose representation is generated for the candidate pose using a transformation model and the candidate pose. The system determines a second probability score for the pose representation based on a pose-representation probability model. The system selects the candidate pose to represent a pose of the body based on at least the first and second probability scores.

Abstract: In one embodiment, a method includes a system accessing an image and generating a feature map using a first neural network. The system identifies a plurality of regions of interest in the feature map. A plurality of regional feature maps may be generated for the plurality of regions of interest, respectively. Using a second neural network, the system may detect at least one regional feature map in the plurality of regional feature maps that corresponds to a person depicted in the image, and generate a target region definition associated with a location of the person using the regional feature map. Based on the target region definition associated with the location of the person, a target regional feature map may be generated by sampling the feature map for the image. The system may process the target regional feature map to generate a keypoint mask and an instance segmentation mask.

Abstract: In one embodiment, a system accesses pose probability models for predetermined parts of a body depicted in an image. Each of the pose probability models is configured for determining a probability of the associated predetermined body part being at a location in the image. The system determines a candidate pose that is defined by a set of coordinates representing candidate locations of the predetermined body parts. The system further determines a first probability score for the candidate pose based on the pose probability models and the set of coordinates of the candidate pose. A pose representation is generated for the candidate pose using a transformation model and the candidate pose. The system determines a second probability score for the pose representation based on a pose-representation probability model. The system selects the candidate pose to represent a pose of the body based on at least the first and second probability scores.

Abstract: In one embodiment, a system may access an image and generate a feature map for the image using a neural network. The system may identify regions of interest in the feature map. Regional feature maps may be generated for the regions of interest, respectively. Each of the regional feature maps has a first, a second, and a third dimension. The system may generate a first combined regional feature map by combining the regional feature maps. The combined regional feature map has a first, a second, and a third dimension. The system may generate a second combined regional feature map by processing the first combined regional feature map using one or more convolutional layers. The system may generate, for each of the regions of interest, information associated with an object instance based on a portion of the second combined regional feature map associated with that region of interest.

Abstract: In one embodiment, a method includes a system accessing an image and generating a feature map using a first neural network. The system identifies a plurality of regions of interest in the feature map. A plurality of regional feature maps may be generated for the plurality of regions of interest, respectively. Using a second neural network, the system may detect at least one regional feature map in the plurality of regional feature maps that corresponds to a person depicted in the image, and generate a target region definition associated with a location of the person using the regional feature map. Based on the target region definition associated with the location of the person, a target regional feature map may be generated by sampling the feature map for the image. The system may process the target regional feature map to generate a keypoint mask and an instance segmentation mask.

Abstract: In one embodiment, a system accesses a probability model associated with an image depicting a body. The probability model includes probability values associated with regions of the image and each probability value represents a probability of the associated region of the image containing a particular body part. The system selects a subset (e.g., 3) of the probability values based on a comparison of the probability values. For each selected probability value, the system identifies surrounding probability values surrounding the selected probability value and computes a probabilistic maximum based on the selected probability value and the surrounding probability values. Each probabilistic maximum is associated with a location within the regions associated with the selected probability value and the surrounding probability values.