Object detection using image alignment for autonomous machine applications

A method comprising:

• converting a first frame and a second frame to aligned images corresponding to a common image plane based at least on rectifying at least one of the first frame or the second frame to the common image plane;
• blending pixel values of aligned pixels between the aligned images to compute difference values indicating differences between the pixel values across the aligned pixels;
• generating a disparity image including third pixels having disparity values corresponding to the difference values and indicating the differences between the pixel values across the aligned pixels;
• combining one or more subsets of the third pixels of the disparity image having the differences above a first threshold value into one or more groups of the third pixels based at least on similarities between the disparity values within the one or more groups;
• and
• generating a bounding shape for at least one of the one or more groups of the third pixels.

A system comprising:

• one or more hardware processing units to execute operations comprising:
• converting a first frame and a second frame to aligned images corresponding to a common image plane based at least on rectifying at least one of the first frame or the second frame to the common image plane;
• blending pixel values of aligned pixels between the aligned images to compute difference values indicating differences between the pixel values across the aligned pixels;
• generating a disparity image including third pixels having disparity values corresponding to the difference values and indicating the differences between the pixel values across the aligned pixels;
• combining one or more subsets of the third pixels of the disparity image having the differences above a first threshold value into one or more groups of the third pixels based at least on similarities between the disparity values within the one or more groups;
• and
• generating a bounding shape for at least one of the one or more groups of the third pixels.