Abstract: A system and method includes determining uncertainty estimation in an object detection deep neural network (DNN) by retrieving a calibration dataset from a validation dataset that includes scores associated with all classes in an image, including a background (BG) class, determining background ground truth boxes in the calibration dataset by comparing ground truth boxes with detection boxes generated by the object detection DNN using an intersection over union (IoU) threshold, correcting for class imbalance between ground truth boxes and background ground truth boxes in a ground truth class by updating the ground truth class to include a number of background ground truth boxes based on a number of ground truth boxes in the ground truth class, estimating uncertainty of the object detection DNN based on the class imbalance correction, and updating output data sets of the object detection DNN based on the class imbalance correction.

Abstract: A dual variational autoencoder-generative adversarial network (VAE-GAN) is trained to transform a real video sequence and a simulated video sequence by inputting the real video data into a real video decoder and a real video encoder and inputting the simulated video data into a synthetic video encoder and a synthetic video decoder. Real loss functions and simulated loss functions are determined based on output from a real video discriminator and a simulated video discriminator, respectively. The real loss functions are backpropagated through the real video encoder and the real video decoder to train the real video encoder and the real video decoder based on the real loss functions. The synthetic loss functions are backpropagated through the synthetic video encoder and the synthetic video decoder to train the synthetic video encoder and the synthetic video decoder based on the synthetic loss functions.

Abstract: A computer includes a processor and a memory storing instructions executable by the processor to receive a dataset of images; extract feature data from the images; optimize a number of clusters into which the images are classified based on the feature data; for each cluster, optimize a number of subclusters into which the images in that cluster are classified; determine a metric indicating a bias of the dataset toward at least one of the clusters or subclusters based on the number of clusters, the numbers of subclusters, distances between the respective clusters, and distances between the respective subclusters; and after determining the metric, train a machine-learning program using a training set constructed from the clusters and the subclusters.

Abstract: A computer includes a processor and a memory, and the memory stores instructions executable by the processor to jointly train a geometric NeRF multilayer perceptron (MLP) and a color NeRF MLP to model a scene using an occupancy grid map, camera data of the scene from a camera, and lidar data of the scene from a lidar; supervise the geometric NeRF MLP with the lidar data during the joint training; and supervise the color NeRF MLP with the camera data during the joint training. The geometric NeRF MLP is a neural radiance field modeling a geometry of the scene, and the color NeRF MLP is a neural radiance field modeling colors of the scene.

Abstract: Image data are input to a machine learning program. The machine learning program is trained with a virtual boundary model based on a distance between a host vehicle and a target object and a loss function based on a real-world physical model. An identification of a threat object is output from the machine learning program. A subsystem of the host vehicle is actuated based on the identification of the threat object.

Abstract: A computer that includes a processor and a memory can predict future status of one or more moving objects by acquiring a plurality of video frames with a sensor included in a device, inputting the plurality of video frames to a first deep neural network to determine one or more objects included in the plurality of video frames, and inputting the objects to a second deep neural network to determine object features and full frame features.

Abstract: A computer includes a processor and a memory storing instructions executable by the processor to receive a plurality of first images of an environment in a first lighting condition, classify pixels of the first images into categories, mask the pixels belonging to at least one of the categories from the first images, generate a three-dimensional representation of the environment based on the masked first images, and generate a second image of the environment in a second lighting condition based on the three-dimensional representation and on a first one of the first images.

Abstract: A computer includes a processor and a memory storing instructions executable by the processor to receive a first image of a scene in a first lighting condition, generate a three-dimensional representation of the scene based on the first image, and generate a second image of the scene in a second lighting condition based on the three-dimensional representation and on the first image. The first image is an only image of the scene used for generating the three-dimensional representation. The first image is an only image of the scene used for generating the second image.

Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to input a fisheye image to a vector quantized variational autoencoder. The vector quantized variational autoencoder can encode the fisheye image to first latent variables based on an encoder. The vector quantized variational autoencoder can quantize the first latent variables to generate second latent variables based on a dictionary of embeddings. The vector quantized variational autoencoder can decode the second latent variables to a rectified rectilinear image using a decoder and output the rectified rectilinear image.

Abstract: A trailer angle identification system includes an imaging device configured to capture an image. A controller is configured to receive steering angle data corresponding to a steering angle of a vehicle, receive vehicle speed data corresponding to a vehicle speed, and estimate an angle of the trailer relative to the vehicle by processing the image, the steering angle data, and the vehicle speed data in at least one neural network.

Abstract: This disclosure describes systems and methods for improved training data acquisition. An example method may include sending, by a processor, an indication for a user to capture data relating to a first area of interest using a first mobile device. The example method may also include determining, by the processor, that first data captured by the first mobile device would fail to satisfy a quality requirement. The example method may also include causing, by the processor, to present an indication through the first mobile device to the user to adjust the first mobile device. The example method may also include determining, by the processor, that second data captured by the first mobile device after being adjusted would satisfy the quality requirement. The example method may also include receiving, by the processor, the second data from the first mobile device.

Abstract: A computer includes a processor and a memory storing instructions executable by the processor to receive a plurality of first images of an environment in a first lighting condition, classify pixels of the first images into categories, mask the pixels belonging to at least one of the categories from the first images, generate a three-dimensional representation of the environment based on the masked first images, and generate a second image of the environment in a second lighting condition based on the three-dimensional representation and on a first one of the first images.

Abstract: A computer includes a processor and a memory storing instructions executable by the processor to receive a first image of a scene in a first lighting condition, generate a three-dimensional representation of the scene based on the first image, and generate a second image of the scene in a second lighting condition based on the three-dimensional representation and on the first image. The first image is an only image of the scene used for generating the three-dimensional representation. The first image is an only image of the scene used for generating the second image.

Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to determine a second convolutional neural network (CNN) training dataset by determining an underrepresented object configuration and an underrepresented noise factor corresponding to an object in a first CNN training dataset, generate one or more simulated images including the object corresponding to the underrepresented object configuration in the first CNN training dataset by inputting ground truth data corresponding to the object into a photorealistic rendering engine and generate one or more synthetic images including the object corresponding to the underrepresented noise factor in the first CNN training dataset by processing the simulated images with a generative adversarial network (GAN) to determine a second CNN training dataset. The instructions can include further instructions to train a CNN to using the first and the second CNN training datasets.

Abstract: Image data are input to a machine learning program. The machine learning program is trained with a virtual boundary model based on a distance between a host vehicle and a target object and a loss function based on a real-world physical model. An identification of a threat object is output from the machine learning program. A subsystem of the host vehicle is actuated based on the identification of the threat object.

Abstract: The present disclosure discloses a system and a method that includes receiving, at a decoder, a latent representation of an image having a first domain, and generating a reconstructed image having a second domain, wherein the reconstructed image is generated based on the latent representation.

Abstract: A training system for a neural network system and method of training is disclosed. The method may comprise: receiving, from a sensor, an image frame captured while an operator is controlling a vehicle; using an eye-tracking system associated with the sensor, monitoring the eyes of the operator to determine eyeball gaze data; determining, from the image frame, a plurality of pedestrians; and iteratively training the neural network system to determine, from among the plurality of pedestrians, the one or more target pedestrians using the eyeball gaze data and an answer dataset that is based on the eyeball gaze data, wherein the determined one or more target pedestrians have a relatively-higher probability of collision with the vehicle than a remainder of the plurality of pedestrians.

Abstract: The present disclosure discloses a system and a method. In an example implementation, the system and the method generate, at a first encoder neural network, an encoded representation of image features of an image received from a vehicle sensor of a vehicle. The system and method can also generate, at a second encoder neural network, an encoded representation of map tile features and generate, at the decoder neural network, a semantically segmented map tile based on the encoded representation of image features, the encoded representation of map tile features, and Global Positioning System (GPS) coordinates of the vehicle. The semantically segmented map tile includes a location of the vehicle and detected objects depicted within the image with respect to the vehicle.

Abstract: A first computer can operate a first instance of a neural network, receive a first data set input to the first instance of the neural network, determine a first calibration parameter for the neural network in the first instance of the neural network based on the first data set, and send the first calibration parameter to a server computer. A second computer can operate a second instance of the neural network, receive a second data set input to the second instance of the neural network, determine a second calibration parameter for the neural network in the second instance of the neural network based on the second data set, and send the second calibration parameter to the server computer. A server computer can aggregate the first and second calibration parameters to update a model of the neural network and update the neural network model for the first and second instances of the neural network at the first and second computers based on the aggregated first and second calibration parameters.

Abstract: A dual variational autoencoder-generative adversarial network (VAE-GAN) is trained to transform a real video sequence and a simulated video sequence by inputting the real video data into a real video decoder and a real video encoder and inputting the simulated video data into a synthetic video encoder and a synthetic video decoder. Real loss functions and simulated loss functions are determined based on output from a real video discriminator and a simulated video discriminator, respectively. The real loss functions are backpropagated through the real video encoder and the real video decoder to train the real video encoder and the real video decoder based on the real loss functions. The synthetic loss functions are backpropagated through the synthetic video encoder and the synthetic video decoder to train the synthetic video encoder and the synthetic video decoder based on the synthetic loss functions.