A computer-implemented method is provided that includes causing an aerial vehicle to scan an environment in a predesignated pattern, such that a first set of images are captured. The method further includes detecting an emergency scene in the first set of images of the environment. The method further includes determining locations at which the aerial vehicle is to capture a second set of images of the emergency scene in the environment. The method further includes causing the aerial vehicle to acquire the second set of images at the locations. The method further includes determining selected images of the second set of images focused on the emergency scene. The method further includes extracting the selected images from the second set of images, the selected images comprising a representation of the emergency scene.