Article: Self-supervised extraction of perspective cues from hand-drawn sketches in environmental design Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.87 pp.96 - 124 Abstract: In the hand-painted deduction in the early stage of environmental design, the 2D concept sketch is easy to cause the 3D space skeleton clues contained in the bottom layer to be extremely chaotic and difficult to analyse. This paper proposes a self-supervised learning framework based on geometric consistency, which is used to automatically extract perspective cues directly from environmental design hand-drawn drafts. The technical mechanism proposed in this paper can directly utilise the inherent orthogonal patterns and vanishing point distribution within the drawings to autonomously establish spatial constraints. Experiments show that the framework proposed in this paper achieves an average accuracy rate of 88.5% in extracting perspective cues, achieving a 4.2% performance improvement compared to the current mainstream benchmark methods, and controlling the parsing time of a single drawing within 1.2 seconds. This method effectively bridges the technical gap between conceptual hand-drawn art and precise computer-aided modelling. Inderscience Publishers - linking academia, business and industry through research

Title: Self-supervised extraction of perspective cues from hand-drawn sketches in environmental design

Authors: Ning Han; Lifang Zhang

Addresses: School of Design and Art, Shandong Huayu University of Technology, Dezhou, 253034, China ' School of Information Engineering, Shandong Huayu University of Technology, Dezhou, 253034, China

Abstract: In the hand-painted deduction in the early stage of environmental design, the 2D concept sketch is easy to cause the 3D space skeleton clues contained in the bottom layer to be extremely chaotic and difficult to analyse. This paper proposes a self-supervised learning framework based on geometric consistency, which is used to automatically extract perspective cues directly from environmental design hand-drawn drafts. The technical mechanism proposed in this paper can directly utilise the inherent orthogonal patterns and vanishing point distribution within the drawings to autonomously establish spatial constraints. Experiments show that the framework proposed in this paper achieves an average accuracy rate of 88.5% in extracting perspective cues, achieving a 4.2% performance improvement compared to the current mainstream benchmark methods, and controlling the parsing time of a single drawing within 1.2 seconds. This method effectively bridges the technical gap between conceptual hand-drawn art and precise computer-aided modelling.

Keywords: environmental design; perspective cues; self-supervised learning; sketch analysis; intelligent design assistance.

DOI: 10.1504/IJICT.2026.155636

International Journal of Information and Communication Technology, 2026 Vol.27 No.87, pp.96 - 124

Received: 10 Apr 2026
Accepted: 13 May 2026
Published online: 07 Aug 2026 *

Title: Self-supervised extraction of perspective cues from hand-drawn sketches in environmental design

Keep up-to-date