Recognition system, information processing apparatus, and information processing method

An information processing apparatus, comprising:

• an interface to receive image data;
• a storage device which stores a commodity file including identification information for a plurality of commodities, feature values for the plurality of commodities, and a predetermined recognition classification for each commodity in the plurality indicating the commodity is one of a first commodity type or a second commodity type, the first commodity type being commodities to be recognized using a code symbol associated with the commodity in the commodity file and the second commodity type being commodities to be recognized using a feature value associated with the commodity in the commodity file;
• and
• a processor which is configured to:
• determine whether a code symbol can be detected in the image data,
• detect an article in the image data if a code symbol cannot be detected in the image data
• extract a feature value of the detected article from the image data,
• calculate similarities between the stored feature values in the commodity file for commodities of the first commodity type and the second commodity type and the extracted feature value,
• recognize commodities in the commodity file as candidates corresponding to the detected article based on the calculated similarities,
• exclude commodities of the first commodity type from the candidates based on the predetermined recognition classification included in the commodity file, and
• select a commodity from the candidates as a recognized commodity based on calculated similarity provided that the calculated similarity for the commodity is above a threshold value.

An information processing method for an information processing apparatus having an interface to receive image data, a storage device which stores a commodity file including identification information for a plurality of commodities, feature values for the plurality of commodities, and a predetermined recognition classification for each commodity in the plurality indicating the commodity is one of a first commodity type or a second commodity type, the first commodity type being commodities to be recognized using a code symbol associated with the commodity in the commodity file and the second commodity type being commodities to be recognized using a feature value associated with the commodity in the commodity file, and a processor, the method comprising:

• determining whether a code symbol can be detected in the image data,
• detecting an article in the image data if a code symbol cannot be detected in the image data,
• extracting a feature value of the detected article from the image data,
• calculating similarities between the stored feature values in the commodity file for commodities of the first commodity type and the second commodity type and the extracted feature value,
• recognizing commodities in the commodity file as candidates corresponding to the detected article based on the calculated similarities,
• excluding commodities of the first commodity type from the candidates based on the predetermined recognition classification included in the commodity file, and
• selecting a commodity from the candidates as a recognized commodity based on calculated similarity provided that the calculated similarity for the commodity is above a threshold value.