Abstract: A video system and method for person search includes video cameras for capturing video images, a display device, and a computer system. The computer system including a deep learning network to determine person images, from among the video images, matching a target query person. The deep learning network having a person detection branch, a person re-identification branch, and an attention-aware relation mixer connected to the person detection branch and to the person re-identification branch. The attention-aware relation mixer including a relation mixer having a spatial and channel mixer that performs spatial attention followed by spatial mixing (tokenized multi-layered perceptron) and channel attention followed by channel mixing (channel multi-layered perceptron), and a joint spatio-channel attention layer that utilizes 3D attention weights to modulate 3D spatio-channel region of interest features and aggregate the features with output of the relation mixer.

Abstract: An apparatus, computer readable storage medium and method of generating a diverse set of images from few-shot images, includes a parameter input receiving values for control parameters to control an extent to which each reference image impacts a newly generated image. The apparatus involves an image generation deep learning network for generating an image for each of the values for the control parameters. The deep learning network has an encoder, a transformer-based fusion block, and a decoder. The transformer-based fusion block includes a mapping network that computes meta-weights from features extracted from the reference images and the control parameters, and a cross-attention block to generate modulation weights based on the meta-weights. An output displays high-quality and diverse images generated based on the values for the control parameter.

Abstract: A system and computer readable storage medium for automated handwriting generation, including a text input device for inputting a text query having at least one textual word string, an image input device for inputting a handwriting sample with characters in a writing style of a user, and a computer implemented deep learning transformer model including an encoder network and a decoder network in which each are a hybrid of convolution and multi-head self-attention networks. The encoder produces a sequence of style feature embeddings from the input handwriting sample. The decoder takes the sequence of style feature embeddings in order to convert the at least one textual word string into a generated handwritten image having substantially same writing style as the handwriting sample. An output device to output the generated handwriting image.

Abstract: This disclosure relates to improved techniques for performing computer vision functions, including common object detection and instance segmentation. The techniques described herein utilize neural network architectures to perform these functions in various types of images, such as natural images, UAV images, satellite images, and other images. The neural network architecture can include a dense location regression network that performs object localization and segmentation functions, at least in part, by generating offset information for multiple sub-regions of candidate object proposals, and utilizing this dense offset information to derive final predictions for locations of target objects. The neural network architecture also can include a discriminative region-of-interest (RoI) pooling network that performs classification of the localized objects, at least in part, by sampling various sub-regions of candidate proposals and performing adaptive weighting to obtain discriminative features.

Abstract: This disclosure relates to improved techniques for performing computer vision functions, including common object detection and instance segmentation. The techniques described herein utilize neural network architectures to perform these functions in various types of images, such as natural images, UAV images, satellite images, and other images. The neural network architecture can include a dense location regression network that performs object localization and segmentation functions, at least in part, by generating offset information for multiple sub-regions of candidate object proposals, and utilizing this dense offset information to derive final predictions for locations of target objects. The neural network architecture also can include a discriminative region-of-interest (Rol) pooling network that performs classification of the localized objects, at least in part, by sampling various sub-regions of candidate proposals and performing adaptive weighting to obtain discriminative features.

Abstract: This disclosure relates to improved techniques for performing computer vision functions including common object counting and instance segmentation. The techniques described herein utilize a neural network architecture to perform these functions. The neural network architecture can be trained using image-level supervision techniques that utilize a loss function to jointly train an image classification branch and a density branch of the neural network architecture. The neural network architecture constructs per-category density maps that can be used to generate analysis information comprising global object counts and locations of objects in images.