Data scientists Brian Kenji Iwana and Seiichi Uchida at Kyushu University in Japan have created a way for a neural network to read a book by its cover.
The computer program uses book cover data to identify which genre the title belongs in. The two researchers uploaded 137,788 book covers from Amazon and tagged the covers with the book’s category among 20 possible genres. Then the researchers trained a neural network to recognize the genre by looking at the cover image. Technology Review has more:
The algorithm listed the correct genre in its top 3 choices over 40 percent of the time and found the exact genre more than 20 percent of the time. That’s significantly better than chance. “This shows that classification of book cover designs is possible, although a very difficult task,” say Iwana and Uchida.
Cookbooks with photos of food and travel books were easiest to detect.