Year: 2024
Type: conference-paper
Abstract: AI models are being used and further developed daily. In graphic design, one of the most common uses is background removal from product images. This paper analyzes the effectiveness of background removal for common products by varying the depth of field. Tests were conducted using two AI background removal models on products of different shapes and textures, with both uniform and structured backgrounds. The results indicate that a shallow depth of field can be beneficial for background removal when the background is not uniform. Conversely, for a uniform background, a larger depth of field can aid AI, especially when the product is larger in size
Authors: Josip Bota, Rebeka Šćulac
Institution: University of Zagreb
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4403644033