Uncertainty modeling for fine-tuned implicit functions
Not on the map. Why?
Abstract (arXiv)
Implicit functions such as Neural Radiance Fields (NeRFs), occupancy networks, and signed distance functions (SDFs) have become pivotal in computer vision for reconstructing detailed object shapes from sparse views. Achieving optimal performance with these models can be challenging due to the extreme sparsity of inputs and distribution shifts induced by data corruptions. To this end, large, noise-free synthetic datasets can serve as shape priors to help models fill in gaps, but the resulting reconstructions must be approached with caution. Uncertainty estimation is crucial for assessing the quality of these reconstructions, particularly in identifying areas where the model is uncertain about the parts it has inferred from the prior. In this paper, we introduce Dropsembles, a novel method for uncertainty estimation in tuned implicit functions. We demonstrate the efficacy of our approach through a series of experiments, starting with toy examples and progressing to a real-world scenario. Specifically, we train a Convolutional Occupancy Network on synthetic anatomical data and test it on low-resolution MRI segmentations of the lumbar spine. Our results show that Dropsembles achieve the accuracy and calibration levels of deep ensembles but with significantly less computational cost.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 56 of its 72 references that are in the map. First posted on arXiv 17 Jun 2024.
Builds on
56 of 72 references.
References · 56
- 2017On Calibration of Modern Neural Networkscited by 4,860
- 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
- 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
- 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
- 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,698
- 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,682
- 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
- 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,214
- 2020Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domainscited by 2,170
- 2018Occupancy Networks: Learning 3D Reconstruction in Function Spacecited by 2,037
- 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,842
- 2020NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collectionscited by 1,230
38 more
- 2021Robust fine-tuning of zero-shot modelscited by 785
- 2020Convolutional Occupancy Networkscited by 760
- 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 735
- 2021Depth-supervised NeRF: Fewer Views and Faster Training for Freecited by 732
- 2020Implicit Geometric Regularization for Learning Shapescited by 707
- 2019Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervisioncited by 699
- 2020Deep Reinforcement Learning for Autonomous Driving: A Surveycited by 698
- 2021A Survey of Uncertainty in Deep Neural Networkscited by 675
- 2021UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstructioncited by 643
- 2021BARF: Bundle-Adjusting Neural Radiance Fieldscited by 641
- 2021RegNeRF: Regularizing Neural Radiance Fields for View Synthesis from Sparse Inputscited by 614
- 2018Predictive Uncertainty Estimation via Prior Networkscited by 555
- 2022MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstructioncited by 467
- 2021Putting NeRF on a Diet: Semantically Consistent Few-Shot View Synthesiscited by 464
- 2019Likelihood Ratios for Out-of-Distribution Detectioncited by 451
- 2021Towards Out-Of-Distribution Generalization: A Surveycited by 431
- 2020Fine-Tuning Pretrained Language Models: Weight Initializations, Data Orders, and Early Stoppingcited by 377
- 2021Dense Depth Priors for Neural Radiance Fields from Sparse Input Viewscited by 301
- 2022SDFusion: Multimodal 3D Shape Completion, Reconstruction, and Generationcited by 287
- 2022SPARF: Neural Radiance Fields from Sparse and Noisy Posescited by 195
- 2021InfoNeRF: Ray Entropy Minimization for Few-Shot Neural Volume Renderingcited by 180
- 2020Posterior Network: Uncertainty Estimation without OOD Samples via Density-Based Pseudo-Countscited by 163
- 2021Modulated Periodic Activations for Generalizable Local Functional Representationscited by 140
- 2022Uncertainty Modeling for Out-of-Distribution Generalizationcited by 122
- 2021Epistemic Neural Networkscited by 120
- 2022Plex: Towards Reliability using Pretrained Large Model Extensionscited by 102
- 2022RelPose: Predicting Probabilistic Relative Rotation for Single Objects in the Wildcited by 102
- 2022COIN++: Neural Compression Across Modalitiescited by 81
- 2021Stochastic Neural Radiance Fields: Quantifying Uncertainty in Implicit 3D Representationscited by 80
- 2022Conditional-Flow NeRF: Accurate 3D Modelling with Reliable Uncertainty Quantificationcited by 71
- 2021On the Practicality of Deterministic Epistemic Uncertaintycited by 62
- 2022AutoRF: Learning 3D Object Radiance Fields from Single View Observationscited by 59
- 2023A Review of Uncertainty Estimation and its Application in Medical Imagingcited by 59
- 2022Assaying Out-Of-Distribution Generalization in Transfer Learningcited by 53
- 2021Wide-Baseline Relative Camera Pose Estimation with Directional Learningcited by 49
- 2024Uncertainty quantification in fine-tuned LLMs using LoRA ensemblescited by 35
- 2023Single-subject Multi-contrast MRI Super-resolution via Implicit Neural Representationscited by 32
- 2021Few-shot Keypoint Detection with Uncertainty Learning for Unseen Speciescited by 27
50 most cited of 56 All 56 (JSON)
Built on by
Citing papers per year: 2024: 1.
1 paper.