Virtual

0
0 ratings - Please login to submit your rating.

Abstract 

This dataset is a valuable resource for those working on virtually rendered scenes and includes comprehensive ground truth data. It features a wide array of images generated under various lighting conditions, specifically designed for tasks such as illumination estimation, scene relighting, and object insertion. Each image is accompanied by precise ground truth information, providing an accurate reference for evaluating and improving algorithms. The dataset includes scenes illuminated by different light sources and angles, ensuring a rich set of examples for realistic lighting simulations. By utilizing this dataset, users can train and validate their algorithms to enhance the accuracy of virtually rendered illumination, thereby improving the visual quality and realism of their projects. The inclusion of ground truth data makes this dataset particularly valuable for a wide range of applications in computer vision, graphics, and augmented reality, supporting both academic research and practical implementations.

Instructions: 

This dataset can be used for validation purposes in case of 3D object insertion because in this dataset ground truth images are also present and named with the letter 'M'. For example, image 'I1' has ground truth 'M1'.

Comments

If you use this dataset, please cite the following article:

 

S. Jamil, K. Amnuayrotchanachinda, and M. A. Amare, "LightDepthMagic: An Advanced Deep Learning and Computer Vision Framework for Realistic 3D Object Embedding in RGB Images," 2024 International Conference on Digital Image Computing: Techniques and Applications (DICTA), Perth, Australia, 2024, pp. 374-381. [DOI: 10.1109/DICTA63115.2024.00062]

 

Submitted by SONAIN JAMIL on Mon, 02/17/2025 - 08:17