New breakthrough in AI dressing up: After training with 1 million photos, the accuracy of deconstructing and reconstructing clothing is 95.7%

IT House reported on November 28 that a scientific research team recently developed the Sewformer AI system. After training 1 million clothing photos, has the ability to identify and reconstruct the characters' clothing in the pictures.

The Sewformer AI system can analyze the single picture data input by the user, deconstruct the clothing worn by the characters in the picture, successfully find out the various parts and gather them together for reconstruction, with an accuracy rate of up to 95.7%, so it can make the clothing generated by AI more reality

Researchers say the most important potential use of this system is virtual reality/metaverse, which can map and construct the current clothing situation of the headset wearer in the digital world, and expand into paid revenue models such as shopping malls.

Of course, there are other potentials for clothing made with artificial intelligence. For example, you can take a photo of someone wearing a certain type of clothing, and then use artificial intelligence to recreate the clothing and make new designs based on it. Regarding this experiment, relevant research has been provided on the preprint server arXiv

IT House provided a paper with a reference address: https://arxiv.org/abs/2311.04218

The above is the detailed content of New breakthrough in AI dressing up: After training with 1 million photos, the accuracy of deconstructing and reconstructing clothing is 95.7%. For more information, please follow other related articles on the PHP Chinese website!