Method for body re-identification using a machine learning model, and method for training a machine learning model for body re-identification
OwnerLan WANG, Lei YUinventors
Life of this patent
Estimating the term…
Abstract
A method for training a machine learning model for body re-identification, where the machine learning model includes a feature map extraction module, an attention weighting mask generating module and a classifying module. The training method includes at least one iteration including: a first training phase which includes processing a first training dataset by the machine learning model and updating the feature extraction module, and a second training phase which includes processing a second training dataset, different from the first training dataset, by the machine learning model and updating the attention weighting mask generating module.
Claims16
A method for training a machine learning model for body re-identification,
- the model comprising a feature map extraction module, an attention weighting mask generating module, and a classifying module, the method comprising at least one iteration comprising a first training phase updating the feature extraction module of the machine learning model based on a first dataset comprising a query image, a target image and a first presence indicator,
- wherein the at least one iteration comprises a second training phase, using a different dataset comprising a training query image, a training target image and a second presence indicator, updating the attention weighting mask generating module based on an evaluation of whether a body in the training query image is present in the training target image.
(canceled)
Dependencies are inferred from the claim text.