Short papers
Up to 4 pages, excluding references and supplementary materials; may introduce original but unfinished research or technical reports using open-source frameworks.
Computer Vision/CORE Unranked
E2E3D 2026
End-to-End 3D Learning (2nd)
Jun 03 2026Official conference site Site reachable
The E2E3D 2026 workshop, held in conjunction with CVPR 2026 in Denver, focuses on end-to-end 3D learning — unified, data-driven systems that map raw images or video directly to 3D representations for real-time applications in robotics, extended reality, and scientific imaging. It aims to replace brittle, multi-stage pipelines with trainable architectures that integrate modeling, inference, and optimization, leveraging large-scale video data, cross-modal learning, and efficient deployment on edge devices.
Key deadlines
Verified September 9, 2026Full paper
May 10, 2026
AoE
Conference timeline
Submission and decisions
Full paperKey deadline
May 10, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Long papers
Up to 8 pages, excluding references and supplementary materials.
Research areas in scope
01
Modeling and learning
Unified architectures mapping pixels to 3D with minimal postprocessingPretraining that embeds geometric priors (scale, viewpoint, occlusion)World models and vision-language-action models using spatial memoryDifferentiable rendering and physics for shape, appearance, and motion
02
Data and pretraining
Open scale data sources (long video, multi-view, multi-sensor logs)Cross-modal alignment using 2D and image-text corporaAuto annotation with quality control and reproducible protocolsData governance (licensing, privacy, safety for 3D assets)
03
Systems, evaluation, and impact
Real-time and edge deployment on robots and mobile devicesHolistic metrics (accuracy, latency, memory, energy)Robustness and safety for open-world generalization and failure analysisApplications in autonomous driving, XR, industrial and scientific imaging, and mapping
Policies worth checking twice
- All submissions must follow the CVPR 2026 LaTeX template format.
- Submissions are non-archival.
- Dual submission is allowed where outside venue rules permit.
- Accepted papers will be presented as posters.
- Best Demo evaluation occurs during the poster/demo session.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026