Authentication and identification of physical objects using fixed-size representation

A method for generating a fixed-size representation of a surface based on microstructural features for identification and authentication of a physical object, the method comprising:

• extracting a plurality of microstructural features from a plurality of surfaces of the physical object using a feature extractor;
• determining an input sequence of tokens, wherein a token from the input sequence of tokens characterize the plurality of microstructural features of the surface of the physical object;
• augmenting the input sequence of tokens with an additional token, wherein the additional token indicates a start or an end of the input sequence of tokens;
• iteratively processing the input sequence of tokens augmented with the additional token using an attention mechanism, wherein:
• contextual information is exchanged between the tokens of the input sequence of tokens, and
• the attention mechanism is trained on a training set of data from the processing of the input sequence of tokens to extract the fixed-size representation of the surface of the physical object from a set of arbitrary size representations containing local microstructural features extracted from the surface;
• determining an output of the processing of the input sequence of tokens augmented with the additional token as the fixed-size representation of the surface;
• and
• determining a status of the physical object based on the fixed-size representation of the surface of the physical object.

A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause a data processing apparatus to perform operations, for generating a fixed-size representation of a surface based on microstructural features for identification and authentication of a physical object, the operations comprising:

• extracting a plurality of microstructural features from a plurality of surfaces of the physical object using a feature extractor;
• determining an input sequence of tokens, wherein a token from the input sequence of tokens characterize the plurality of microstructural features of the surface of the physical object;
• augmenting the input sequence of tokens with an additional token, wherein the additional token indicates a start or an end of the input sequence of tokens;
• iteratively processing the input sequence of tokens augmented with the additional token using an attention mechanism, wherein:
• contextual information is exchanged between the tokens of the input sequence of tokens, and
• the attention mechanism is trained on a training set of data from processing of the input sequence of tokens to extract the fixed-size representation of the surface of the physical object from a set of arbitrary size representations containing local microstructural features extracted from the surface;
• determining an output of the processing of the input sequence of tokens augmented with the additional token as the fixed-size representation of the surface;
• and
• determining a status of the physical object based on the fixed-size representation of the surface of the physical object.