Perspective Sketch

Perspective sketching research focuses on bridging the gap between human-drawn sketches and digital 3D models, aiming to improve the efficiency and intuitiveness of design and reasoning processes. Current research explores novel deep learning architectures, including diffusion models and occupancy networks, to translate perspective sketches into 3D representations, often leveraging complementary information from other views or textual descriptions. This work has implications for various fields, from architectural design and urban planning (facilitating rapid 3D model generation from sketches) to improving human-computer interaction and multimodal AI systems (by enabling more natural and effective sketch-based interfaces for problem-solving). The development of large-scale, diverse sketch datasets is also a key focus, enabling better training and evaluation of these models.