3D Assembly

DeepCAD is a CAD dataset consisting of 179,133 models and their CAD construction sequences. It can be used to train generative models of 3D shapes.

Breaking Bad is a large-scale dataset of fractured objects. The dataset contains around 10k meshes from PartNet and Thingi10k. For each mesh, 20 fracture modes are pre-computed and then simulate 80 fractures from them, resulting in a total of 1M breakdown patterns. This dataset serves as a benchmark that enables the study of fractured object reassembly and presents new challenges for geometric shape understanding. Source: Breaking Bad: A Dataset for Geometric Fracture and Reassembly Image Source: https://arxiv.org/pdf/2210.11463.pdf

Our dataset consists of over 1000 fractured frescoes. The RePAIR stands as a realistic computational challenge for methods for 2D and 3D puzzle solving, and serves as a benchmark that enables the study of fractured object reassembly and presents new challenges for geometric shape understanding. Please visit our website for more dataset information, access to source code scripts and for an interactive gallery viewing of the dataset samples.

3D design file repository for the Stickbug Robot a 6 armed holonomic precision pollination robot

The RAD (Randomly Assembled Object Construction) dataset is a synthetic 3D LEGO dataset designed for the task of Sequential Brick Assembly (SBA). Here are the key characteristics and details: High-level explanation: Consists of 3D objects built with 2x4 LEGO bricks. Objects are randomly assembled following predefined connection configurations. Includes variations like RAD-S and RAD with different numbers of bricks per object. Provides ground truth labels for assembly actions. Includes multi-view images, sequential assembly actions, 3D voxel, conjunction graph, LEGO ldr file, etc. Motivations and content summary: Created to provide a large-scale synthetic dataset for training and evaluating SBA models. Allows for controlled experimentation with varying object complexity. RAD-1k: 1000 objects with around 15 bricks each. RAD-S: 10000 objects built with around 20 bricks each. RAD: 10000 objects built with around 60 bricks each. Expands on previous datasets by allowing more flexible brick connections, including bottom-to-top assembly. Potential use cases: Pre-training models for sequential brick assembly tasks. Evaluating the scalability of assembly algorithms with increasing object complexity. Studying transfer learning from synthetic to real-world assembly tasks. Benchmarking performance of different SBA approaches. Investigating generalization capabilities of assembly models to different object structures. The RAD dataset provides a valuable resource for researchers working on 3D object assembly, particularly in developing and testing algorithms that can handle increasingly complex structures built from simple primitives.