FlowNet3D: Learning Scene Flow in 3D Point Clouds

Charles R. Qi; Leonidas J. Guibas; Xingyu Liu

RECORD

FlowNet3D: Learning Scene Flow in 3D Point Clouds

Xingyu Liu, Charles R. Qi, Leonidas J. Guibas

Computer Vision and Pattern Recognition · Submitted 4 Jun 2018

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Trained on synthetic data only, our network successfully generalizes to real scans, outperforming various baselines and showing competitive results to the prior art.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 5,548 input tokens.

What the abstract itself states, as probabilities: claims first 0.10 · states limitation 0.32 · human participants 0.04 · controlled comparison 0.84 · randomized assignment 0.01 · states quantitative result 0.05 · compares with prior results 0.97 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v3. Results and the main claim are underlined; show every role.

Many applications in robotics and human-computer interaction can benefit from understanding 3D motion of points in a dynamic environment, widely noted as scene flow. While most previous methods focus on stereo and RGB-D images as input, few try to estimate scene flow directly from point clouds. In this work, we propose a novel deep neural network named $FlowNet3D$ that learns scene flow from point clouds in an end-to-end fashion. Our network simultaneously learns deep hierarchical features of point clouds and flow embeddings that represent point motions, supported by two newly proposed learning layers for point sets. We evaluate the network on both challenging synthetic data from FlyingThings3D and real Lidar scans from KITTI. Trained on synthetic data only, our network successfully generalizes to real scans, outperforming various baselines and showing competitive results to the prior art. We also demonstrate two applications of our scene flow output (scan registration and motion segmentation) to show its potential wide use cases.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#	sentence	role	the authors' own work	states the main conclusion	strength
s0	Many applications in robotics and human-computer interaction can benefit from understanding 3D motion of points in a dynamic environment, widely noted as scene flow.	background 1.00	0.04	0.02	asserted 0.68
s1	While most previous methods focus on stereo and RGB-D images as input, few try to estimate scene flow directly from point clouds.	gap 0.96	0.08	0.02	asserted 0.99
s2	In this work, we propose a novel deep neural network named $FlowNet3D$ that learns scene flow from point clouds in an end-to-end fashion.	contribution 0.95	0.95	0.12	asserted 0.99
s3	Our network simultaneously learns deep hierarchical features of point clouds and flow embeddings that represent point motions, supported by two newly proposed learning layers for point sets.	method 0.93	0.93	0.06	asserted 1.00
s4	We evaluate the network on both challenging synthetic data from FlyingThings3D and real Lidar scans from KITTI.	method 0.91	0.96	0.04	asserted 0.98
s5	Trained on synthetic data only, our network successfully generalizes to real scans, outperforming various baselines and showing competitive results to the prior art.	result 1.00	0.97	0.69	asserted 1.00
s6	We also demonstrate two applications of our scene flow output (scan registration and motion segmentation) to show its potential wide use cases.	contribution 0.77	0.95	0.07	asserted 0.98

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.89.

What was examined, and what was not

sources read

arXiv listing abstract v3

not examined

full text
figures and tables
references and citing sentences
relations to other papers

levels run

P abstract

not run

S sections
T body sentences
C citations
R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper

Original abstract

Many applications in robotics and human-computer interaction can benefit from understanding 3D motion of points in a dynamic environment, widely noted as scene flow. While most previous methods focus on stereo and RGB-D images as input, few try to estimate scene flow directly from point clouds. In this work, we propose a novel deep neural network named $FlowNet3D$ that learns scene flow from point clouds in an end-to-end fashion. Our network simultaneously learns deep hierarchical features of point clouds and flow embeddings that represent point motions, supported by two newly proposed learning layers for point sets. We evaluate the network on both challenging synthetic data from FlyingThings3D and real Lidar scans from KITTI. Trained on synthetic data only, our network successfully generalizes to real scans, outperforming various baselines and showing competitive results to the prior art. We also demonstrate two applications of our scene flow output (scan registration and motion segmentation) to show its potential wide use cases.

Human review

How Pith checks papers → · Full audit of this paper →

References

How the literature uses it

7 arXiv papers cite this in 16 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 7 sentences without a confident label

Uses its method or tool 2 papers

Since disparity is not given for the test set, we evaluate on all 142 scenes in the training set with publicly available raw 3D data, following [ref] .

HPLFlowNet: Hierarchical Permutohedral Lattice FlowNet for Scene Flow Estimation on Large-scale Point Clouds · Generalization results on real-world data · 0.84
Following [ref] we report the average precision (AP) metric for dynamic/static segmentation by predicted velocity value.

Any Motion Detector: Learning Class-agnostic Scene Dynamics from a Sequence of LiDAR Point Clouds · Evaluation Datasets and Metrics · 0.95

Compares with it 2 papers

FlowNet3D [ref] : the state-of-the-art for scene flow estimation with point cloud inputs.

HPLFlowNet: Hierarchical Permutohedral Lattice FlowNet for Scene Flow Estimation on Large-scale Point Clouds · Results on FlyingThings3D · 0.89
We compare our method with FlowNet3D [ref] and ICP [ref] .

Any Motion Detector: Learning Class-agnostic Scene Dynamics from a Sequence of LiDAR Point Clouds · Baselines · 0.99

Cites it as background 5 papers

Classical and learned scene-flow methods estimate dense 3D motion between observed frames, while recent trackers recover long-range correspondences through camera motion, occlusion, and nonrigid deformation [ref] .

MotionForesight: Re-purposing Video Models for Future 3D Scene-Flow Prediction · Related Work · 0.97
Some algorithms train neural network models from unstructured LiDAR point clouds [ref] .

LiDAR-Flow: Dense Scene Flow Estimation from Sparse LiDAR and Stereo Images · RELATED WORK · 0.96
Generative learning using exact-likelihood methods based on invertible transformations has had remarkable success in domains like accurately representing images [ref] , audio [ref] , 3D point cloud data [ref] as well as applications in physics [ref] .

Equivariant Flows: sampling configurations for multi-body systems with symmetric energies · Introduction · 0.98

What it cites · 6 sentences without a confident label

Uses the method or tool of 1 paper

We therefore follow a recently proposed PointNet++ architecture [ref] , a translation-invariant network that learns hierarchical features.

PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space · Hierarchical Point Cloud Feature Learning · 1.00

Cites as background 1 paper

[ref] jointly estimate object rigid motions and segment them based on their motions.

Motion-based Object Segmentation based on Dense RGB-D Scene Flow · Related Work · 0.82

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.