Information processing method and information processing system

An information processing system, comprising:

• a processor;
• and
• a memory that stores a program,
• wherein when the program is executed by the processor, the program causes the processor to execute:
• obtaining sensing data pieces from a sensor;
• determining a synthesis region in a first sensing data piece in which a recognition target data piece is to be synthesized with the first sensing data piece;
• generating a composite data piece by synthesizing the recognition target data piece into the synthesis region of the first sensing data piece, wherein the recognition target data piece has same or similar perceived characteristics perceived by a human sensory system as the sensing data piece;
• obtaining recognition result data by providing the composite data piece to a neural network which has been trained using machine learning to recognize a recognition target, wherein the recognition target data piece of the composite data piece includes a data piece of a first object corresponding to the recognition target, wherein the recognition result data indicates a recognition result of the first object recognized by the neural network;
• judging whether a recognition accuracy of the first object is lower than a threshold, wherein the recognition accuracy of the first object is calculated based on the recognition result data and reference data including at least the synthesis region;
• determining two or more training data pieces to update the neural network based on the composite data piece when it is judged that the recognition accuracy of the first object is lower than a threshold;
• and
• updating the neural network using the two or more training data pieces,
• wherein, the determining determines, as the two or more training data pieces, (i) a first corresponding data piece being a second sensing data piece which is selected from among the sensing data pieces obtained from the sensor and which includes a data piece of a second object having same or similar characteristics as the first object, in the composite data piece, of which the recognition accuracy is determined to be lower than the threshold and (ii) a second corresponding data piece which is generated from a third sensing data piece among the sensing data pieces obtained from the sensor and which includes the data piece of the second object.