List of available models

Below is a list and a brief explanation about the models currently available on PTLFlow.

List of models

CCMR

[source code]

• Paper: CCMR: High Resolution Optical Flow Estimation via Coarse-to-Fine Context-Guided Motion Reasoning - https://arxiv.org/abs/2311.02661

• Reference code: https://github.com/cv-stuttgart/CCMR

• Model names: ccmr, ccmr_p

CRAFT

[source code]

• Paper: CRAFT: Cross-Attentional Flow Transformers for Robust Optical Flow - https://arxiv.org/abs/2203.16896

• Reference code: https://github.com/askerlee/craft

• Model names: craft

CSFlow

[source code]

• Paper: CSFlow: Learning optical flow via cross strip correlation for autonomous driving - https://arxiv.org/abs/2202.00909

• Reference code: https://github.com/MasterHow/CSFlow

• Model names: csflow

DICL-Flow

[source code]

• Paper: Displacement-Invariant Matching Cost Learning for Accurate Optical Flow Estimation - https://arxiv.org/abs/2010.14851

• Reference code: https://github.com/jytime/DICL-Flow

• Model names: dicl

DIP

[source code]

• Paper: DIP: Deep Inverse Patchmatch for High-Resolution Optical Flow - https://arxiv.org/abs/2204.00330

• Reference code: https://github.com/zihuazheng/DIP

• Model names: dip

DPFlow

[source code]

• Paper: DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework - https://arxiv.org/abs/2503.14880

• Model names: dpflow

FastFlownet

[source code]

• Paper: FastFlowNet: A Lightweight Network for Fast Optical Flow Estimation - https://arxiv.org/abs/2103.04524

• Reference code: https://github.com/ltkong218/FastFlowNet

• Model names: fastflownet

Flow1D

[source code]

• Paper: High-Resolution Optical Flow from 1D Attention and Correlation - https://arxiv.org/abs/2104.13918

• Reference code: https://github.com/haofeixu/flow1d

• Model names: flow1d

Flow-Anything

[source code]

• Paper: Flow-Anything: Learning Real-World Optical Flow Estimation from Large-Scale Single-view Images - https://arxiv.org/abs/2506.07740

• Reference code: https://github.com/Sharpiless/Flow-Anything

• Model names: flow_anything

FlowFormer

[source code]

• Paper: FlowFormer: A Transformer Architecture for Optical Flow - https://arxiv.org/abs/2203.16194

• Reference code: https://github.com/drinkingcoder/FlowFormer-Official

• Model names: flowformer

FlowFormer++

[source code]

• Paper: Flowformer++: Masked cost volume autoencoding for pretraining optical flow estimation - https://arxiv.org/abs/2303.01237

• Reference code: https://github.com/XiaoyuShi97/FlowFormerPlusPlus

• Model names: flowformer_pp

Flownet

[source code]

• Papers:

  • FlowNet: Learning Optical Flow with Convolutional Networks - https://arxiv.org/abs/1504.06852

  • FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks - https://arxiv.org/abs/1612.01925

• Reference code: https://github.com/NVIDIA/flownet2-pytorch

• Model names: flownets, flownetc, flownet2, flownetcs, flownetcss, flownetsd

FlowSeek

[source code]

• Paper: FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases - https://arxiv.org/abs/2509.05297

• Reference code: https://github.com/mattpoggi/flowseek

• Model names: flowseek_t, flowseek_m

GMA

[source code]

• Paper: Learning to Estimate Hidden Motions with Global Motion Aggregation - https://arxiv.org/abs/2104.02409

• Reference code: https://github.com/zacjiang/GMA

• Model names: gma

GMFlow

[source code]

• Paper: GMFlow: Learning Optical Flow via Global Matching - https://arxiv.org/abs/2111.13680

• Reference code: https://github.com/haofeixu/gmflow

• Model names: gmflow, gmflow_refine

GMFlow+, UniMatch

[source code]

• Paper: Unifying Flow, Stereo and Depth Estimation - https://arxiv.org/abs/2211.05783

• Reference code: https://github.com/autonomousvision/unimatch

• Model names: gmflow_p, gmflow_p_sc2, gmflow_p_sc2_ref6, unimatch, unimatch_sc2, unimatch_sc2_ref6

GMFlowNet

[source code]

• Paper: Global Matching with Overlapping Attention for Optical Flow Estimation - https://arxiv.org/abs/2203.11335

• Reference code: https://github.com/xiaofeng94/GMFlowNet

• Model names: gmflownet, gmflownet_mix

HD3

[source code]

• Paper: Hierarchical Discrete Distribution Decomposition for Match Density Estimation - https://arxiv.org/abs/1812.06264

• Reference code: https://github.com/ucbdrive/hd3

• Model names: hd3, hd3_ctxt

IRR

[source code]

• Paper: Iterative Residual Refinement for Joint Optical Flow and Occlusion Estimation - https://arxiv.org/abs/1904.05290

• Reference code: https://github.com/visinf/irr

• Model names: irr_pwc, irr_pwcnet, irr_pwcnet_irr

LCV

[source code]

• Paper: Learnable Cost Volume Using the Cayley Representation - https://arxiv.org/abs/2007.11431

• Reference code: https://github.com/Prinsphield/LCV

• Model names: lcv_raft, lcv_raft_small

LiteFlowNet

[source code]

• Paper: LiteFlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation - https://arxiv.org/abs/1805.07036

• Reference code: https://github.com/twhui/LiteFlowNet

• Model name: liteflownet

LiteFlowNet2

[source code]

• Paper: A Lightweight Optical Flow CNN - Revisiting Data Fidelity and Regularization - https://ieeexplore.ieee.org/document/9018073

• Reference code: https://github.com/twhui/LiteFlowNet2

• Model names: liteflownet2, liteflownet2_pseudoreg

LiteFlowNet3

[source code]

• Paper: LiteFlowNet3: Resolving Correspondence Ambiguity for More Accurate Optical Flow Estimation - https://arxiv.org/abs/2007.09319

• Reference code: https://github.com/twhui/LiteFlowNet3

• Model names: liteflownet3, liteflownet3_pseudoreg, liteflownet3s, liteflownet3s_pseudoreg

LLA-Flow

[source code]

• Paper: LLA-Flow: A Lightweight Local Aggregation on Cost Volume for Optical Flow Estimation - https://arxiv.org/abs/2304.08101

• Reference code: https://github.com/mansang127/LLA-Flow

• Model names: llaflow, llaflow_raft

MaskFlownet

[source code]

• Paper: MaskFlownet: Asymmetric Feature Matching with Learnable Occlusion Mask - https://arxiv.org/abs/2003.10955

• Reference code: https://github.com/cattaneod/MaskFlownet-Pytorch

• Model names: maskflownet, maskflownet_s

MatchFlow

[source code]

• Paper: Rethinking Optical Flow from Geometric Matching Consistent Perspective - https://arxiv.org/abs/2303.08384

• Reference code: https://github.com/DQiaole/MatchFlow

• Model names: matchflow, matchflow_raft

MemFlow

[source code]

• Paper: MemFlow: Optical Flow Estimation and Prediction with Memory - https://arxiv.org/abs/2404.04808

• Reference code: https://github.com/DQiaole/MemFlow

• Model names: memflow, memflow_t

MEMFOF

[source code]

• Paper: MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation - https://arxiv.org/abs/2506.23151

• Reference code: https://github.com/msu-video-group/memfof

• Model name: memfof

MS-RAFT+

[source code]

• Paper: High-Resolution Multi-Scale RAFT - https://arxiv.org/abs/2210.16900

• Reference code: https://github.com/cv-stuttgart/MS_RAFT_plus

• Model names: ms_raft_p

NeuFlow v1

[source code]

• Paper: NeuFlow: Real-time, High-accuracy Optical Flow Estimation on Robots Using Edge Devices - https://arxiv.org/abs/2403.10425

• Reference code: https://github.com/neufieldrobotics/neuflow

• Model names: neuflow

NeuFlow v2

[source code]

• Paper: NeuFlow v2: Push High-Efficiency Optical Flow To the Limit - https://arxiv.org/abs/2408.10161

• Reference code: https://github.com/neufieldrobotics/NeuFlow_v2

• Model names: neuflow2

PWCNet

[source code]

• Paper: PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume - https://arxiv.org/abs/1709.02371

• Reference code: https://github.com/NVlabs/PWC-Net

• Model names: pwcnet, pwcnet_nodc

RAFT

[source code]

• Paper: RAFT: Recurrent All-Pairs Field Transforms for Optical Flow - https://arxiv.org/abs/2003.12039

• Reference code: https://github.com/princeton-vl/RAFT

• Model names: raft, raft_small

RAPIDFlow

[source code]

• Paper: RAPIDFlow: Recurrent Adaptable Pyramids with Iterative Decoding for Efficient Optical Flow Estimation - https://hmorimitsu.com/publication/2024-icra-rapidflow/

• Model names: rapidflow, rapidflow_it1, rapidflow_it2, rapidflow_it3, rapidflow_it6, rapidflow_it12

ReCoVEr

[source code]

• Paper: Removing Cost Volumes from Optical Flow Estimators - https://arxiv.org/abs/2510.13317

• Reference code: https://github.com/visinf/recover

• Model names: recover_cx, recover_mn, recover_rn

RPKNet

[source code]

• Paper: Recurrent Partial Kernel Network for Efficient Optical Flow Estimation - https://hmorimitsu.com/publication/2024-aaai-rpknet

• Model names: rpknet

ScopeFlow

[source code]

• Paper: ScopeFlow: Dynamic Scene Scoping for Optical Flow - https://arxiv.org/abs/2002.10770

• Reference code: https://github.com/avirambh/ScopeFlow

• Model names: scopeflow

SCV

[source code]

• Paper: Learning Optical Flow from a Few Matches - https://arxiv.org/abs/2104.02166

• Reference code: https://github.com/zacjiang/SCV

• Model names: scv4, scv8

SEA-RAFT

[source code]

• Paper: SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical Flow - https://arxiv.org/abs/2405.14793

• Reference code: https://github.com/princeton-vl/SEA-RAFT

• Model names: sea_raft, sea_raft_s, sea_raft_m, sea_raft_l

SeparableFlow

[source code]

• Paper: Separable Flow: Learning Motion Cost Volumes for Optical Flow Estimation - https://openaccess.thecvf.com/content/ICCV2021/papers/Zhang_Separable_Flow_Learning_Motion_Cost_Volumes_for_Optical_Flow_Estimation_ICCV_2021_paper.pdf

• Reference code: https://github.com/feihuzhang/SeparableFlow

• Model names: separableflow

SKFlow

[source code]

• Paper: SKFlow: Learning Optical Flow with Super Kernels - https://arxiv.org/abs/2205.14623

• Reference code: https://github.com/littlespray/SKFlow

• Model names: skflow

SplatFlow

[source code]

• Paper: SplatFlow: Learning Multi-frame Optical Flow via Splatting - https://arxiv.org/abs/2306.08887

• Reference code: https://github.com/wwsource/SplatFLow

• Model names: splatflow

STaRFlow

[source code]

• Paper: STaRFlow: A SpatioTemporal Recurrent Cell for Lightweight Multi-Frame Optical Flow Estimation - https://arxiv.org/abs/2007.05481

• Reference code: https://github.com/pgodet/star_flow

• Model names: starflow

StreamFlow

[source code]

• Paper: StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences - https://arxiv.org/abs/2311.17099

• Model names: streamflow

VCN

[source code]

• Paper: Volumetric Correspondence Networks for Optical Flow - https://papers.nips.cc/paper/2019/file/bbf94b34eb32268ada57a3be5062fe7d-Paper.pdf

• Reference code: https://github.com/gengshan-y/VCN

• Model names: vcn, vcn_small

VideoFlow

[source code]

• Paper: Videoflow: Exploiting temporal cues for multi-frame optical flow estimation - https://arxiv.org/abs/2303.08340

• Reference code: https://github.com/XiaoyuShi97/VideoFlow

• Model names: videoflow_bof, videoflow_mof

WAFT

[source code]

• Paper: WAFT: Warping-Alone Field Transforms for Optical Flow - https://arxiv.org/abs/2506.21526

• Reference code: https://github.com/princeton-vl/WAFT

• Model name: waft_dav2_a1, waft_dav2_a2, waft_dinov3_a2, waft_twins_a2