LAS&T is a large-scale, highly diverse dataset for shape, texture, and material recognition and retrieval, with 700,000 images based on real-world shapes and textures.
Scripts Additional resources Paper
The LAS&T Dataset aims to test the most basic aspect of vision in the most general way. Mainly the ability to identify any shape, texture, and material in any setting and environment, without being limited to specific types or classes of objects, materials, and environments. For shapes, this means identifying and retrieving any shape in 2D or 3D with every element of the shape changed between images, including the material and texture, orientation, size, and environment. For textures and materials, the goal is to recognize the same texture or material when appearing on different objects, environments, and light conditions. The dataset relies on shapes, textures, and materials extracted from real-world images, leading to an almost unlimited quantity and diversity of real-world natural patterns. Each section of the dataset (shapes, and textures), contains 3D parts that rely on physics-based scenes with realistic light materials and object simulation and abstract 2D parts. In addition, a real-world images benchmark for 3D shapes recognition is also supplied.
3D shape recognition and retrieval: The goal is to identify the same 3D shape in different images with different materials, orientations, environments, and lighting.
3D_Shape_large_synthetic_set (Set1, Set2) maximum variability for training/testing
3D_shapes_Real_world_benchmark 130 real world images for testing
3D_shjapes_Different_sub_tests various of sub tests (See paper)
2D shapes recognition and retrieval. The goal is to identify the same 2D shape in different images with different textures, orientations, and backgrounds. Real and synthetic images.
All shapes and textures are extracted from natural images.
A repository of 350,000 extracted shapes is also supplied.
2D_shapes_large_synthetic_set (Set1, Set2, Set3, Set 4) maximum variability for training/testing
2D_shapes_Different_sub_tests various of sub tests (See paper)
365K_Complesx_2D_Shapes Giant set of 350,000 2D shapes saved as PNG images
30K Semantic shapes Masks of semantic objects
3D Materials recognition and retrieval. The goal is to identify the same 3D materials in different images when appear on different objects and environments and lighting.
3D_Materials_large_synthetic_set (Set1, Set2, Set3) maximum variability for training/testing
3D_materials_Different_sub_tests various of sub tests (See paper)
2D texture recognition and retrieval. The goal is to identify the same texture in different images when appear on different shapes and backgrounds.
2D_Textures_large_synthetic_set (Set1, Set2, Set3, Set4) maximum variability for training/testing
2D_Textures_Different_sub_tests various of sub tests (See paper)
3D Shapes Synthetic image generation scripts
3D Materials Synthetic image generation scripts
2D Shapes extraction from natural images script
2D Shapes synthetic image generation scripts
2D Textures synthetic image generation scripts
Testing + Evaluation scripts (For LVLM and Human testing and generating multi-choice test images)
Other useful assets:
Vastexture: Giant repository of textures and PBR materials
Objaverse: Giant Repository of 3D Objects
Paper:
Shape and Texture Recognition in Large Vision-Language Models