Fast Feature Field (F3): A Predictive Representation of Events
Not on the map. Why?
Abstract (arXiv)
This paper develops a mathematical argument and algorithms for building representations of data from event-based cameras, that we call Fast Feature Field (F3). We learn this representation by predicting future events from past events and show that it preserves scene structure and motion information. F3 exploits the sparsity of event data and is robust to noise and variations in event rates. It can be computed efficiently using ideas from multi-resolution hash encoding and deep sets - achieving 120 Hz at HD and 440 Hz at VGA resolutions. F3 represents events within a contiguous spatiotemporal volume as a multi-channel image, enabling a range of downstream tasks. We obtain state-of-the-art performance on optical flow estimation, semantic segmentation, and monocular metric depth estimation, on data from three robotic platforms (a car, a quadruped robot and a flying platform), across different lighting conditions (daytime, nighttime), environments (indoors, outdoors, urban, as well as off-road) and dynamic vision sensors (resolutions and event rates). Our implementations can predict these tasks at 25-75 Hz at HD resolution.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 66 of its 106 references that are in the map. First posted on arXiv 29 Sep 2025.
Builds on
66 of 106 references.
References · 66
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
- 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
- 2016Layer Normalizationcited by 5,669
- 2017On Calibration of Modern Neural Networkscited by 4,960
- 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,444
- 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,723
- 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,940
- 2020RAFT: Recurrent All-Pairs Field Transforms for Optical Flowcited by 2,415
- 2019Towards Robust Monocular Depth Estimation: Mixing Datasets for Zero-shot Cross-dataset Transfercited by 1,495
- 2024Depth Anything V2cited by 1,422
- 2017Deep Setscited by 1,421
38 more
- 2022VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Trainingcited by 1,322
- 2014Depth Map Prediction from a Single Image using a Multi-Scale Deep Networkcited by 1,274
- 20194D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networkscited by 1,208
- 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
- 2019Event-based Vision: A Surveycited by 1,016
- 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 809
- 2023ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoderscited by 712
- 2020TartanAir: A Dataset to Push the Limits of Visual SLAMcited by 506
- 2022InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutionscited by 482
- 2020Calibrating Deep Neural Networks using Focal Losscited by 357
- 2019High Speed and High Dynamic Range Video with an Event Cameracited by 345
- 2018Unsupervised Event-based Learning of Optical Flow, Depth, and Egomotioncited by 307
- 2021DSEC: A Stereo Event Camera Dataset for Driving Scenarioscited by 303
- 2018The Multi Vehicle Stereo Event Camera Dataset: An Event Camera Dataset for 3D Perceptioncited by 254
- 2018EV-FlowNet: Self-Supervised Optical Flow Estimation for Event-based Camerascited by 224
- 2018HATS: Histograms of Averaged Time Surfaces for Robust Event-based Object Classificationcited by 218
- 2017Emergence of Invariance and Disentanglement in Deep Representationscited by 215
- 2018Event-based Vision meets Deep Learning on Steering Prediction for Self-driving Carscited by 200
- 2020Hierarchical Multi-Scale Attention for Semantic Segmentationcited by 183
- 2021E-RAFT: Dense Optical Flow from Event Camerascited by 152
- 2017Ultimate SLAM? Combining Events, Images, and IMU for Robust Visual SLAM in HDR and High Speed Scenarioscited by 150
- 2018Event-based Moving Object Detection and Trackingcited by 127
- 2018Asynchronous, Photometric Feature Tracking using Events and Framescited by 118
- 2022AEGNN: Asynchronous Event-based Graph Neural Networkscited by 118
- 2018EV-SegNet: Semantic Segmentation for Event-based Camerascited by 93
- 2021Combining Events and Frames using Recurrent Asynchronous Multimodal Networks for Monocular Depth Predictioncited by 90
- 2023Deep Learning for Event-based Vision: A Comprehensive Survey and Benchmarkscited by 90
- 2020Learning Monocular Dense Depth from Eventscited by 85
- 2022ESS: Learning Event-based Semantic Segmentation from Still Imagescited by 82
- 2023From Chaos Comes Order: Ordering Event Representations for Object Recognition and Detectioncited by 52
- 2019EVDodgeNet: Deep Dynamic Obstacle Dodging with Event Camerascited by 46
- 2023Taming Contrast Maximization for Learning Sequential, Low-latency, Event-based Optical Flowcited by 39
- 2023TorchSparse++: Efficient Training and Inference Framework for Sparse Convolution on GPUscited by 36
- 2022Event Collapse in Contrast Maximization Frameworkscited by 33
- 2021Dual Transfer Learning for Event-based End-task Prediction via Pluggable Event to Image Translationcited by 32
- 2022EVIMO2: An Event Camera Dataset for Motion Segmentation, Optical Flow, Structure from Motion, and Visual Inertial Odometry in Indoor Scenes with Monocular or Stereo Algorithmscited by 31
- 2022Lightweight Event-based Optical Flow Estimation via Iterative Deblurringcited by 31
- 2023Fully neuromorphic vision and control for autonomous drone flightcited by 30
50 most cited of 66 All 66 (JSON)
Built on by
Citing papers per year: 2026: 3.
3 papers.