Online self-supervised learning depth estimation method based on binocular endoscope

The invention provides an online self-supervised learning depth estimation method and system based on a binocular endoscope, and relates to the technical field of depth estimation. The method comprises the following steps: acquiring a binocular endoscope image, and extracting multi-scale features of a current frame image by adopting an encoder network of a current binocular depth estimation network; fusing the multi-scale features by adopting a decoder network of the current binocular depth estimation network to obtain parallax of each pixel point in the current frame image; according to internal and external parameters of the camera, converting the parallax into depth and outputting the depth as a result of the current frame image; under the condition that an external truth value is not introduced, parameters of a current binocular depth estimation network are updated by utilizing self-supervision loss for depth estimation of a next frame of image. According to the method, the similarity of continuous frames