Method and device for training neural network for image recognition

The invention discloses a method and equipment for training a neural network for image recognition. The method includes: determining first trunk feature data corresponding to each input data by applying two or more sets of input data, which are images of the same scene, to a first neural network model, respectively; determining second trunk feature data corresponding to each piece of input data by applying two or more sets of input data to a second neural network model, respectively; determining projection-based first embedded data and discard-based first view data from the first trunk feature data; determining projection-based second embedded data and discard-based second view data from the second trunk feature data; and training any one or both of the first neural network model and the second neural network model based on a loss determined based on a combination of any two or more of the first embedded data, the first view data, the second embedded data, the second view data, and the embedded data clusterin