Hierarchical SfM paper at ECCV 2026 scales feedforward 3D reconstruction with VGGT-Ω
fdellaert · x · 2026-09-10
A team presented "Hierarchical Structure-from-Motion Scales Feedforward Reconstruction" at the ECCV 2026 SfM workshop, combining 3D foundation models with geometric optimization: split an image collection into overlapping groups, reconstruct each with VGGT-Ω, then progressively merge and refine pieces into a globally consistent 3D model. Interactive demos on the project page show reconstructions of Grand Place, Thanjavur and Gerrard Hall coming together.
More from Research
- ReactVAU at ECCV: slow-fast decoupling cuts MLLM calls for streaming video anomaly understanding — CMHungSteven · 2026-09-10
- ReactVAU at ECCV 2026: fast-slow streaming video understanding cuts heavy MLLM calls — CMHungSteven · 2026-09-10
- Perplexity releases Q2D-Web: a 190M-document retrieval benchmark for agentic RAG — antoine_chaffin · 2026-09-10
- Correction: the SmolVLM data-filtering trick is from DeepSeek's earlier tech report — eliebakouch · 2026-09-10
- DeepSeek used SmolVLM to quality-filter interleaved pretraining data, researcher spots — eliebakouch · 2026-09-10
- SmolVLM used for strict image-text quality scoring in new model tech report — eliebakouch · 2026-09-10