The impact of intermediate video frames reconstruction step on the result of 3D reconstruction of objects

The task of 3D reconstruction of objects is one of the most challenging tasks in computer vision. This is usually caused by the requirement of a lot of input information, such as a large number of images or camera settings. Often the available data for object reconstruction is in the form of a limited number of images from the Internet or frames of a video sequence with a low rate. This paper estimates the performance of such widely used 3D reconstruction methods, as Akool, COLMAP, and NeRF depending on the input data. The obtained results show that the stage of intermediate frame reconstruction allows to improve the quality of 3D reconstruction in the case of a low number of input images.