Abstract: A computer-implemented method for transforming a neural radiance field model is described. A plurality of inputs are provided to a neural radiance field (NeRF) model that represents a 3-dimensional space having a subject, wherein each input of the plurality of inputs includes a location and a view direction and corresponds to respective colors of voxels that represent the 3-dimensional space. A spectral analysis is performed on a plurality of outputs of the NeRF model based on the plurality of inputs, wherein the plurality of outputs include the respective colors of the voxels. Frequency components of the spectral analysis that represent colors for at least some of the voxels are extracted. A sparse volume data structure that represents the 3-dimensional space and the respective colors for the at least some of the voxels is generated.

Abstract: A system configured to improve human presence detection and/or localization by generating aggregate confidence values. The system may aggregate confidence values corresponding to overlapping regions of interest. The system may perform human presence detection by comparing the aggregate confidence values to a universal threshold, with aggregate confidence values above the universal threshold indicating that human presence is detected. The system may use the aggregate confidence values to generate a heatmap, may identify a strongest cluster of pixels in the heatmap and determine a bounding box surrounding the strongest cluster. To distinguish the strongest cluster from a second strongest cluster of pixels, the system may apply a Gaussian kernel with varying threshold values. The system may store the threshold value at which the strongest cluster separates from the second strongest cluster and may use the threshold value to generate the bounding box.

Abstract: Devices and techniques are generally described for segmentation of object image data from background image data. In some examples, the segmentation of object image data may comprise capturing image data comprising color data and depth data. In some examples, the segmentation of object image data may further include separating the depth data into a plurality of clusters of image data, wherein each cluster is associated with a respective range of depth values. In various examples, the segmentation of object image data may comprise selecting a main cluster of image data as corresponding to an object of interest in the image data. In various other examples, the segmentation of object image data may comprise identifying pixels of the main cluster that correspond to the object of interest.

Abstract: Devices and techniques are generally described for segmentation of object image data from background image data. In some examples, the segmentation of object image data may comprise capturing image data comprising color data and depth data. In some examples, the segmentation of object image data may further include separating the depth data into a plurality of clusters of image data, wherein each cluster is associated with a respective range of depth values. In various examples, the segmentation of object image data may comprise selecting a main cluster of image data as corresponding to an object of interest in the image data. In various other examples, the segmentation of object image data may comprise identifying pixels of the main cluster that correspond to the object of interest.

Abstract: A system configured to improve human presence detection and/or localization by generating aggregate confidence values. The system may aggregate confidence values corresponding to overlapping regions of interest. The system may perform human presence detection by comparing the aggregate confidence values to a universal threshold, with aggregate confidence values above the universal threshold indicating that human presence is detected. The system may use the aggregate confidence values to generate a heatmap, may identify a strongest cluster of pixels in the heatmap and determine a bounding box surrounding the strongest cluster. To reduce false positive detections, the system may track the false positive detections as heatmap base data and may remove the heatmap base data from the heatmap. Thus, pixel intensity values corresponding to the false positive detections may be reduced to improve an accuracy of the human presence detection.

Abstract: Devices and techniques are generally described for segmentation of image data using depth data. In various examples, color image data may be received from a digital camera. In some examples, depth image data may be received from a depth sensor. In various examples, the depth image data may be separated into a plurality of clusters of depth image data, wherein each cluster is associated with a respective range of depth values. In some further examples, a determination may be made that a first cluster of image data corresponds to an object of interest, such as a human subject, in the image data. In various examples, pixels of the first cluster may be encoded with foreground indicator data. In some further examples, segmented image data may be generated. The segmented image data may comprise pixels encoded with the foreground indicator data.