Method

DPFlow [DPFlow]
https://github.com/hmorimitsu/ptlflow/tree/main/ptlflow/models/dpflow

Submitted on 21 Oct. 2024 04:10 by
Henrique Morimitsu (University of Science and Technology Beijing)

Running time:0.1 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
DPFlow is an optical flow architecture that uses a
dual-pyramid bidirectional recurrent encoder to
adapt to inputs of multiple sizes. Our method can
be trained only with low-resolution samples and
generalize well to 8K resolution inputs.

We also created the Kubric-NK dataset, featuring
2400 densely annotated samples at four resolutions
ranging from 1K to 8K. We use Kubric-NK to
evaluate multiple existing optical flow methods
and reveal new findings about how high-resolution
inputs affect the predictions.
Parameters:
Please refer to the paper and source code.
Latex Bibtex:
@InProceedings{Morimitsu2025DPFlow,
author = {Morimitsu, Henrique and Zhu, Xiaobin
and Cesar-Jr., Roberto M. and Ji, Xiangyang and Yin,
Xu-Cheng},
booktitle = {CVPR},
title = {{DPFlow}: Adaptive Optical Flow
Estimation with a Dual-Pyramid Framework},
year = {2025},
}

Detailed Results

This page provides detailed results for the method(s) selected. For the first 20 test images, the percentage of erroneous pixels is depicted in the table. We use the error metric described in Object Scene Flow for Autonomous Vehicles (CVPR 2015), which considers a pixel to be correctly estimated if the disparity or flow end-point error is <3px or <5% (for scene flow this criterion needs to be fulfilled for both disparity maps and the flow map). Underneath, the left input image, the estimated results and the error maps are shown (for disp_0/disp_1/flow/scene_flow, respectively). The error map uses the log-color scale described in Object Scene Flow for Autonomous Vehicles (CVPR 2015), depicting correct estimates (<3px or <5% error) in blue and wrong estimates in red color tones. Dark regions in the error images denote the occluded pixels which fall outside the image boundaries. The false color maps of the results are scaled to the largest ground truth disparity values / flow magnitudes.

Test Set Average

Error Fl-bg Fl-fg Fl-all
All / All 3.29 4.93 3.56
All / Est 3.29 4.93 3.56
Noc / All 2.06 2.40 2.12
Noc / Est 2.06 2.40 2.12
This table as LaTeX

Test Image 0

Error Fl-bg Fl-fg Fl-all
All / All 1.39 15.28 3.30
All / Est 1.39 15.28 3.30
Noc / All 1.45 15.28 3.54
Noc / Est 1.45 15.28 3.54
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 1

Error Fl-bg Fl-fg Fl-all
All / All 1.69 7.72 2.36
All / Est 1.69 7.72 2.36
Noc / All 1.57 7.72 2.33
Noc / Est 1.57 7.72 2.33
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 2

Error Fl-bg Fl-fg Fl-all
All / All 2.87 5.45 2.99
All / Est 2.87 5.45 2.99
Noc / All 2.54 5.45 2.71
Noc / Est 2.54 5.45 2.71
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 3

Error Fl-bg Fl-fg Fl-all
All / All 6.31 8.26 6.49
All / Est 6.31 8.26 6.49
Noc / All 3.82 6.15 4.02
Noc / Est 3.82 6.15 4.02
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 4

Error Fl-bg Fl-fg Fl-all
All / All 1.62 5.03 2.18
All / Est 1.62 5.03 2.18
Noc / All 1.29 4.18 1.83
Noc / Est 1.29 4.18 1.83
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 5

Error Fl-bg Fl-fg Fl-all
All / All 2.33 0.17 2.13
All / Est 2.33 0.17 2.13
Noc / All 2.30 0.17 2.08
Noc / Est 2.30 0.17 2.08
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 6

Error Fl-bg Fl-fg Fl-all
All / All 3.94 0.48 3.58
All / Est 3.94 0.48 3.58
Noc / All 2.11 0.48 1.92
Noc / Est 2.11 0.48 1.92
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 7

Error Fl-bg Fl-fg Fl-all
All / All 1.29 1.67 1.36
All / Est 1.29 1.67 1.36
Noc / All 1.29 1.76 1.37
Noc / Est 1.29 1.76 1.37
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 8

Error Fl-bg Fl-fg Fl-all
All / All 0.24 1.38 0.45
All / Est 0.24 1.38 0.45
Noc / All 0.24 1.38 0.45
Noc / Est 0.24 1.38 0.45
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 9

Error Fl-bg Fl-fg Fl-all
All / All 0.62 1.42 0.82
All / Est 0.62 1.42 0.82
Noc / All 0.62 1.42 0.82
Noc / Est 0.62 1.42 0.82
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 10

Error Fl-bg Fl-fg Fl-all
All / All 1.00 2.97 1.45
All / Est 1.00 2.97 1.45
Noc / All 1.03 2.97 1.53
Noc / Est 1.03 2.97 1.53
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 11

Error Fl-bg Fl-fg Fl-all
All / All 1.16 0.95 1.12
All / Est 1.16 0.95 1.12
Noc / All 1.15 0.95 1.11
Noc / Est 1.15 0.95 1.11
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 12

Error Fl-bg Fl-fg Fl-all
All / All 1.61 0.78 1.55
All / Est 1.61 0.78 1.55
Noc / All 1.06 0.78 1.04
Noc / Est 1.06 0.78 1.04
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 13

Error Fl-bg Fl-fg Fl-all
All / All 2.82 1.10 2.61
All / Est 2.82 1.10 2.61
Noc / All 1.28 0.62 1.20
Noc / Est 1.28 0.62 1.20
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 14

Error Fl-bg Fl-fg Fl-all
All / All 2.59 1.35 2.57
All / Est 2.59 1.35 2.57
Noc / All 1.89 1.35 1.88
Noc / Est 1.89 1.35 1.88
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 15

Error Fl-bg Fl-fg Fl-all
All / All 8.38 1.90 7.79
All / Est 8.38 1.90 7.79
Noc / All 5.97 1.90 5.50
Noc / Est 5.97 1.90 5.50
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 16

Error Fl-bg Fl-fg Fl-all
All / All 12.39 3.53 11.09
All / Est 12.39 3.53 11.09
Noc / All 6.58 3.53 6.03
Noc / Est 6.58 3.53 6.03
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 17

Error Fl-bg Fl-fg Fl-all
All / All 3.47 1.59 3.27
All / Est 3.47 1.59 3.27
Noc / All 3.11 1.59 2.92
Noc / Est 3.11 1.59 2.92
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 18

Error Fl-bg Fl-fg Fl-all
All / All 14.13 84.19 47.42
All / Est 14.13 84.19 47.42
Noc / All 9.07 54.95 21.72
Noc / Est 9.07 54.95 21.72
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 19

Error Fl-bg Fl-fg Fl-all
All / All 2.83 2.15 2.75
All / Est 2.83 2.15 2.75
Noc / All 2.59 2.15 2.52
Noc / Est 2.59 2.15 2.52
This table as LaTeX

Input Image

Flow Result

Flow Error




eXTReMe Tracker