Method

FDFlowNet: Fast Optical Flow Estimation using a Deep Lightweight Network [FDFlowNet]


Submitted on 22 Jun. 2020 14:15 by
Lingtong Kong (SJTU)

Running time:0.02 s
Environment:NVIDIA GTX 1080 Ti

Method Description:
Significant progress has been made for estimating
optical flow using deep neural networks. Advanced
deep models achieve accurate flow estimation
often with a considerable computation complexity
and time-consuming training processes. In this
work, we present a lightweight yet effective
model for real-time optical flow estimation,
termed FDFlowNet (fast deep flownet). We achieve
better or similar accuracy on the challenging
KITTI and Sintel benchmarks while being about 2
times faster than PWC-Net. This is achieved by a
carefully-designed structure and newly proposed
components. We first introduce an U-shape network
for constructing multi-scale feature which
benefits upper levels with global receptive field
compared with pyramid network. In each scale, a
partial fully connected structure with dilated
convolution is proposed for flow estimation that
obtains a good balance among speed, accuracy and
number of parameters compared with sequential
connected and dense connected structures.
Experiments demonstrate that our model achieves
state-of-the-art performance while being fast and
lightweight.
Parameters:
FDFlowNet
Latex Bibtex:
@inproceedings{kong2020fdflownet,
author={Lingtong Kong and Jie Yang},
booktitle={IEEE International Conference on
Image Processing (ICIP)},
title={FDFlowNet: Fast Optical Flow Estimation
using a Deep Lightweight Network},
year={2020},}

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 9.31 9.71 9.38
All / Est 9.31 9.71 9.38
Noc / All 5.35 6.62 5.58
Noc / Est 5.35 6.62 5.58
This table as LaTeX

Test Image 0

Error Fl-bg Fl-fg Fl-all
All / All 3.80 25.61 6.80
All / Est 3.80 25.61 6.80
Noc / All 3.44 25.61 6.80
Noc / Est 3.44 25.61 6.80
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 1

Error Fl-bg Fl-fg Fl-all
All / All 3.85 21.95 5.87
All / Est 3.85 21.95 5.87
Noc / All 3.40 21.95 5.69
Noc / Est 3.40 21.95 5.69
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 2

Error Fl-bg Fl-fg Fl-all
All / All 8.34 14.48 8.64
All / Est 8.34 14.48 8.64
Noc / All 6.25 14.48 6.73
Noc / Est 6.25 14.48 6.73
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 3

Error Fl-bg Fl-fg Fl-all
All / All 18.29 16.73 18.15
All / Est 18.29 16.73 18.15
Noc / All 15.04 5.60 14.23
Noc / Est 15.04 5.60 14.23
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 4

Error Fl-bg Fl-fg Fl-all
All / All 6.05 14.19 7.40
All / Est 6.05 14.19 7.40
Noc / All 5.01 12.00 6.32
Noc / Est 5.01 12.00 6.32
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 5

Error Fl-bg Fl-fg Fl-all
All / All 3.92 0.12 3.57
All / Est 3.92 0.12 3.57
Noc / All 3.09 0.12 2.78
Noc / Est 3.09 0.12 2.78
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 6

Error Fl-bg Fl-fg Fl-all
All / All 6.34 0.86 5.76
All / Est 6.34 0.86 5.76
Noc / All 4.01 0.86 3.64
Noc / Est 4.01 0.86 3.64
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 7

Error Fl-bg Fl-fg Fl-all
All / All 2.51 25.37 6.99
All / Est 2.51 25.37 6.99
Noc / All 2.51 21.35 6.03
Noc / Est 2.51 21.35 6.03
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 8

Error Fl-bg Fl-fg Fl-all
All / All 1.19 4.71 1.84
All / Est 1.19 4.71 1.84
Noc / All 1.19 4.71 1.84
Noc / Est 1.19 4.71 1.84
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 9

Error Fl-bg Fl-fg Fl-all
All / All 1.67 11.00 4.05
All / Est 1.67 11.00 4.05
Noc / All 1.67 11.00 4.05
Noc / Est 1.67 11.00 4.05
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 10

Error Fl-bg Fl-fg Fl-all
All / All 2.69 4.18 3.03
All / Est 2.69 4.18 3.03
Noc / All 2.29 4.18 2.78
Noc / Est 2.29 4.18 2.78
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 11

Error Fl-bg Fl-fg Fl-all
All / All 4.01 2.65 3.77
All / Est 4.01 2.65 3.77
Noc / All 3.90 2.62 3.64
Noc / Est 3.90 2.62 3.64
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 12

Error Fl-bg Fl-fg Fl-all
All / All 3.07 4.08 3.14
All / Est 3.07 4.08 3.14
Noc / All 2.23 4.08 2.38
Noc / Est 2.23 4.08 2.38
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 13

Error Fl-bg Fl-fg Fl-all
All / All 4.92 2.60 4.64
All / Est 4.92 2.60 4.64
Noc / All 1.82 1.74 1.81
Noc / Est 1.82 1.74 1.81
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 14

Error Fl-bg Fl-fg Fl-all
All / All 5.35 0.67 5.27
All / Est 5.35 0.67 5.27
Noc / All 2.84 0.67 2.80
Noc / Est 2.84 0.67 2.80
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 15

Error Fl-bg Fl-fg Fl-all
All / All 12.82 5.03 12.12
All / Est 12.82 5.03 12.12
Noc / All 9.14 5.03 8.67
Noc / Est 9.14 5.03 8.67
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 16

Error Fl-bg Fl-fg Fl-all
All / All 17.80 7.96 16.35
All / Est 17.80 7.96 16.35
Noc / All 10.10 7.96 9.71
Noc / Est 10.10 7.96 9.71
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 17

Error Fl-bg Fl-fg Fl-all
All / All 11.28 3.93 10.52
All / Est 11.28 3.93 10.52
Noc / All 7.29 3.93 6.85
Noc / Est 7.29 3.93 6.85
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 18

Error Fl-bg Fl-fg Fl-all
All / All 20.56 100.00 58.30
All / Est 20.56 100.00 58.30
Noc / All 12.39 100.00 36.55
Noc / Est 12.39 100.00 36.55
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 19

Error Fl-bg Fl-fg Fl-all
All / All 8.29 4.49 7.86
All / Est 8.29 4.49 7.86
Noc / All 4.58 4.49 4.57
Noc / Est 4.58 4.49 4.57
This table as LaTeX

Input Image

Flow Result

Flow Error




eXTReMe Tracker