2results about How to "Labeling is simple" patented technology

This invention discloses a training method for a scene text erasure model. A scene text detection dataset is used as the training set for the scene text erasure model. The last classification layer of the baseline model is replaced with two parallel classification layers, thus dividing the entire model into a background restoration branch and a text erasure branch, resulting in the scene text erasure model. The background restoration branch is trained by taking a partially occluded background image as input and predicting the background filling content of text regions and randomly occluded regions. During training, the background image is used as the label for this branch to supervise its learning process. The text erasure branch is trained by taking an input image as input and predicting the background filling content after text regions are erased and restored. During training, the replaced image is used as the pseudo-label for this branch to supervise its learning process. This invention trains the scene text erasure model using only a text detection dataset in a weakly supervised manner.

This disclosure relates to a model training method, apparatus, vehicle, storage medium, and computer program product, belonging to the field of image processing technology. The method includes iteratively executing the following steps to obtain a target network model: inputting a collected image of a storage location into an initial network model to predict the predicted three-dimensional coordinates of a target point in the storage location; converting the predicted three-dimensional coordinates into predicted two-dimensional coordinates; updating the network parameters in the initial network model based on the loss value between the predicted two-dimensional coordinates and the actual two-dimensional coordinates; the actual two-dimensional coordinates are obtained by labeling the target point in the storage location image; if the loss value meets the convergence condition, the initial network model that meets the convergence condition is used as the target network model; the target network model is used to obtain the three-dimensional coordinates of the target point from the storage location image. Using the model training method proposed in this disclosure, the cost of manual labeling can be reduced while obtaining highly accurate predicted three-dimensional coordinates.