DeepJEB++ scales the DeepJEB dataset by an order of magnitude, synthesizing 15,360 jet-engine brackets from 380 SimJEB seed geometries via 2D latent-space interpolation and 3D foundation models. Each sample provides node-level FEA labels: mass, displacement, and von Mises stress fields under four distinct load cases, establishing a benchmark dataset directly applicable to training physics-informed surrogate models and generative engineering design frameworks. 

The DeepWheel dataset is a publicly available resource designed to facilitate data-driven research in wheel design. It includes topology-optimized images, photorealistic renderings, depth estimation outputs, 3D models, and structural analysis data.

This open dataset, meticulously crafted through the synergy of 3D deep learning technology and Finite Element Analysis (FEA) simulations, serves as a robust foundation for various advanced engineering applications. It is particularly suited for 3D data processing and machine learning endeavors demanding substantial data volume.