Abstract: A method of compositional feature representation learning for video understanding is described. The method includes individually processing a sequence of video frames received as an input of a feature map network to generate a plurality of feature maps. The method also includes binding the plurality of feature maps to a fixed set of slot variables using an attention model according to a motion segmentation signal. The method further includes combining slot states corresponding to the fixed set of slot variables into a combined feature map. The method also includes decoding the combined feature map to form a reconstructed sequence of video frames, in which objects discovered in the reconstructed sequence of video frames are identified.

Abstract: A method and system for sensing an obstacle comprises transmitting an electromagnetic signal from a mobile machine to an object. A reflected electromagnetic signal is received from the object to determine a distance between the object and the mobile machine. An image patch is extracted from a region associated with the object. Each image patch comprises coordinates (e.g., three dimensional coordinates) associated with corresponding image data (e.g., pixels). If an object is present, image data may include at least one of object density data and object color data. Object density data is determined based on a statistical measure of variation associated with the image patch. Object color data based on the color of the object detected with brightness normalization. An object is classified or identified based on the determined object density and determined object color data.