Intelligent bulldozer operation identification method based on vehicle-mounted vision and multi-task learning

The invention belongs to the technical field of earthwork machinery intelligent identification, and discloses a bulldozer operation intelligent identification method based on vehicle-mounted vision and multi-task learning. The method comprises the following steps: S1, arranging a plurality of cameras on the bulldozer so as to collect video data when the bulldozer works; for the collected video data, according to the content of soil in front of a shovel blade of the bulldozer, the shoveling state and the shoveling motion are marked, so that a database in which the video data correspond to the shoveling state and the shoveling motion is formed; and S2, constructing a multi-task learning network, training the multi-task learning network by using the database in the S1 so as to obtain an identification model, collecting real-time video data of the bulldozer in real time, and inputting the real-time video data into the identification model so as to obtain the shoveling state and shoveling motion of the bulldozer.