Artificial intelligence finds Chinese imperial artifact
BERLIN, GERMANY, March 16, 2021 /EINPresswire.com/ -- A German startup, Artai.biz, has developed a new service model capable of recognizing historical artifacts in Internet pictures and photographs. The user just needs to photograph an object, and, in a split second, the service will identify the artifact depicted in pictures, illustrations or photographs of historical sources.
In the antique world, an artifact is considered authentic if it has a convincing provenance: the history of its origin, a detailed list of its owners and publications about it in books and magazines. This new service will help to significantly reduce the cost of establishing the provenance of a historical artifact, as well as increase the speed of examination.
The service is based on a neural network pre-trained on over 100 million archival documents, books and magazines. Due to the extensive database used, the search performance results are impressive. As an example, an experiment was conducted with a randomly selected historical artifact from Wikipedia.
The results revealed that a similar image was found in a historical illustration of the early 20th century.
The illustration depicts the Chinese Emperor Guangxu, with an object in the background that looks like the historical artifact used in this sample search. To confirm the accuracy of their hypothesis, the startup team contacted an expert, Jeremiah Jenne (a well-known expert in the field of Chinese history), for help, who confirmed that the horse-shaped image is indeed a historical artifact, which previously belonged to Emperor Guangxu and is now in a private collection. Its estimated cost could range in tens of millions of dollars.
In the antique world, an artifact is considered authentic if it has a convincing provenance: the history of its origin, a detailed list of its owners and publications about it in books and magazines. This new service will help to significantly reduce the cost of establishing the provenance of a historical artifact, as well as increase the speed of examination.
The service is based on a neural network pre-trained on over 100 million archival documents, books and magazines. Due to the extensive database used, the search performance results are impressive. As an example, an experiment was conducted with a randomly selected historical artifact from Wikipedia.
The results revealed that a similar image was found in a historical illustration of the early 20th century.
The illustration depicts the Chinese Emperor Guangxu, with an object in the background that looks like the historical artifact used in this sample search. To confirm the accuracy of their hypothesis, the startup team contacted an expert, Jeremiah Jenne (a well-known expert in the field of Chinese history), for help, who confirmed that the horse-shaped image is indeed a historical artifact, which previously belonged to Emperor Guangxu and is now in a private collection. Its estimated cost could range in tens of millions of dollars.
Jane Weber
Artai.biz
info@artai.biz
