Abstract: A physics-based network model is trained to learn weights such as trapping, detrapping, and/or transport of holes and/or electrons, as well as voltage distribution on a voxel-by-voxel basis throughout a solid-state detector model. The physics-based network may be used to estimate material property variation throughout the voxels. To reduce the number of experimental setups and information needed to train the models, the models may be trained using more easily acquired ground truth. Just the electrode signals or just the free charge data is used to train the model to characterize the solid-state detector. With this reduced data, the detector may be characterized using equivalency, such as combining multiple trapping centers to an equivalent trapping center. Regularization may be used in the loss calculation, such as where just the electrode signals are used, to deal with the reduced data available as ground truth.

Abstract: A system to process imaging data includes an imaging system configured to capture image data and event data of a scene, compress the image data and the event data, and transmit the compressed image data and compressed event data to a host. The host is operatively coupled to the imaging system, and includes a processor configured to perform object detection on the compressed image data and the compressed event data to identify one or more objects. The processor is also configured to perform object tracking on the one or more objects. The processor is also configured to predict one or more regions of interest for subsequent data capture based on the object detection and the object tracking. The processor is further configured to provide the one or more regions of interest to the imaging system to control capture of additional image data and additional event data by the imaging system.

Abstract: A physics-based network model is trained to learn weights such as trapping, detrapping, and/or transport of holes and/or electrons, as well as voltage distribution on a voxel-by-voxel basis throughout a solid-state detector model. The physics-based network may be used to estimate material property variation throughout the voxels. Anode and cathode signals as well as the voltage distribution are relatively strong signals compared to the weaker electron and hole signals. The relatively weaker signals may be limited in range across voxels. In order to expand the range or magnify the effect, the loss function used in training the physics-based neural network may use a weighted combination where the weaker signals are weighted more heavily than stronger signals without substantially reducing the influence of the stronger signals. This improves the inference, resulting in improvement of the accuracy and range of the trained physics-based model.

Abstract: A method of training a neural network modeling physical phenomena of semiconductor material includes receiving plurality of training pairs corresponding to a semiconductor material. Each training pair comprises an input charge to a distinct voxel of the semiconductor material and one or more output signals generated by the distinct voxel in response to the input charge. A neural network is trained using the training pairs. The neural network models the semiconductor material and each voxel is represented in the neural network by a tensor field defined by (i) a location of the voxel within the semiconductor material and (ii) one or more physics-based phenomena within the voxel at the location.

Abstract: A system to process imaging data includes an imaging system configured to capture image data and event data of a scene, compress the image data and the event data, and transmit the compressed image data and compressed event data to a host. The host is operatively coupled to the imaging system, and includes a processor configured to perform object detection on the compressed image data and the compressed event data to identify one or more objects. The processor is also configured to perform object tracking on the one or more objects. The processor is also configured to predict one or more regions of interest for subsequent data capture based on the object detection and the object tracking. The processor is further configured to provide the one or more regions of interest to the imaging system to control capture of additional image data and additional event data by the imaging system.

Abstract: A method of training a neural network modeling physical phenomena of semiconductor material includes receiving plurality of training pairs corresponding to a semiconductor material. Each training pair comprises an input charge to a distinct voxel of the semiconductor material and one or more output signals generated by the distinct voxel in response to the input charge. A neural network is trained using the training pairs. The neural network models the semiconductor material and each voxel is represented in the neural network by a tensor field defined by (i) a location of the voxel within the semiconductor material and (ii) one or more physics-based phenomena within the voxel at the location.

Abstract: For training to and/or estimating location, energy level, and/or time of occurrence of incident radiation on a solid-state detector, a machine-learned model, such as a neural network, performs the inverse problem. An estimate of the location, energy level, and/or time is output by the machine-learned model in response to input of the detected signal (e.g., voltage over time). The estimate may account for material property variation of the solid-state detector in a rapid and easily calculated way, and with a minimal amount of data.