arXiv Science⌕ Search

arXiv subjects

Guangxi Feng

Publications and source records attributed to Guangxi Feng.

2 recordsLinked to original sources

Toward On-Device Markerless 6DoF Tracking of a Physical Manipulative on a Mixed-Reality Headset: A Preliminary Study

AR-Classroom uses physical models and virtual overlays to help learners explore spatial transformations. To align these overlays on a standalone headset, we developed a markerless six-degree-of-freedom(6DoF) tracker for a LEGO airplane on the Meta Quest 3. A compact YOLO model trained on synthetic Unity renders predicts 20 keypoints from dynamic RGB crops resized to 128 by 128 pixels. A perspective-n-point solver estimates the airplane's pose from these predictions and the corresponding 3D landmarks. In a paired evaluation on 1,000 synthetic validation images, cropping guided by ground-truth boxes increased the detection rate on annotated images from 32.5 to 46.2 percent. Mean keypoint F1 increased from 42.9 to 44.6 percent, with lower keypoint precision in the cropped view. In preliminary headset tests, 2 Hz was the highest tested tracking target at which the application maintained approximately 72 frames per second. A headset demonstration illustrates virtual overlay alignment with the physical airplane before and after a manual rotation. Quantitative 6DoF pose accuracy on real headset footage will be measured in future work.

cs.CV↗

Text2VP: Generative AI for Visual Programming and Parametric Modeling

The integration of generative artificial intelligence (AI) into architectural design has advanced significantly, enabling the generation of text, images, and 3D models. However, prior AI applications lack support for text-to-parametric models, essential for generating and optimizing diverse parametric design options. This study introduces Text-to-Visual Programming (Text2VP) GPT, a novel generative AI derived from GPT-4.1, designed to automate graph-based visual programming workflows, parameters, and their interconnections. Text2VP leverages detailed documentation, specific instructions, and example-driven few-shot learning to reflect user intentions accurately and facilitate interactive parameter adjustments. Testing demonstrates Text2VP's capability in generating functional parametric models, although higher complexity models present increased error rates. This research highlights generative AI's potential in visual programming and parametric modeling, laying groundwork for future improvements to manage complex modeling tasks. Ultimately, Text2VP aims to enable designers to easily create and modify parametric models without extensive training in specialized platforms like Grasshopper.

cs.HC↗