Skip to content
/ Video-Frame-Interpolation-Rankings-and-Video-Deblurring-Rankings Public

Rankings include: ABME AdaFNIO ALANET AMT BiT BVFI CDFI CtxSyn DBVI DeMFI DQBC DRVI EAFI EBME EDC EDENVFI EDSC EMA-VFI FGDCN FILM FLAVR HiFI H-VFI IFRNet IQ-VFI JNMR LADDER M2M MA-GCSPA NCM PerVFI PRF ProBoost-Net RIFE RN-VFI SoftSplat SSR ST-MFNet Swin-VFI TDPNet TTVFI UGFI UPR-Net UTI-VFI VFIformer VFIFT VFIMamba VFIT VIDUE VRT

100 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

AIVFI/Video-Frame-Interpolation-Rankings-and-Video-Deblurring-Rankings

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

66 Commits
MoSt-DSA.png
MoSt-DSA.png
 
 
README.md
README.md
 
 
cat.png
cat.png
 
 

Repository files navigation

Video Frame Interpolation Rankings
and Video Deblurring Rankings

Researchers! Please develope joint video deblurring and frame interpolation models, incorporate InterpAny-Clearer in your models and train at least one of your models on Style loss, also called Gram matrix loss (the best perceptual loss function):

FILM - Loss Functions Ablation Source: FILM - Loss Functions Ablation https://film-net.github.io/

MoSt-DSA - Loss Function Comparison

Source: MoSt-DSA - Loss Function Comparison https://arxiv.org/html/2407.07078

List of Rankings

Each ranking includes only the best model for one method.

The rankings exclude all event-based and spike-guided models.

Joint Video Deblurring and Frame Interpolation Rankings

  1. 👑 RBI with real motion blur✔️: LPIPS😍 (no data)
    This will be the King of all rankings. We look forward to ambitious researchers.
  2. RBI with real motion blur✔️: PSNR😞>=28.5dB
  3. Adobe240 (640×352) with synthetic motion blur✖️: LPIPS😍 (no data)
  4. Adobe240 (640×352) with synthetic motion blur✖️: PSNR😞>=33.3dB
  5. Adobe240 (5:8) with synthetic motion blur✖️: LPIPS😍 (no data)
  6. Adobe240 (5:8) with synthetic motion blur✖️: PSNR😞>=25dB

Video Deblurring Rankings

  • (to do)

Video Frame Interpolation Rankings

  1. Vimeo-90K triplet: LPIPS😍(SqueezeNet)<=0.014
  2. Vimeo-90K triplet: LPIPS😍<=0.018
  3. Vimeo-90K triplet: PSNR😞>=36dB
  4. Vimeo-90K septuplet: LPIPS😍<=0.032
  5. Vimeo-90K septuplet: PSNR😞>=36dB

Appendices

RBI with real motion blur✔️: PSNR😞>=28.5dB

RK     Model        PSNR ↑   
{Input fr.}		 Training
dataset		 Official
  repository  		 Practical
model		 VapourSynth
1 Pre-BiT++
CVPR		 31.32 {3}
CVPR		 Pretraining: Adobe240
Training: RBI		 GitHub Stars Request -
2 DeMFI-Netrb(5,3)
ECCV		 29.03 {4}
CVPR		 RBI GitHub Stars - -
3 PRF4 -Large
CVPR
ENH:
TIP		 28.55 {5}
CVPR		 RBI GitHub Stars - -

Back to Top Back to the List of Rankings

Adobe240 (640×352) with synthetic motion blur✖️: PSNR😞>=33.3dB

RK     Model        PSNR ↑   
{Input fr.}		 Originally
announced
or Training
dataset		 Official
  repository  		 Practical
model		 VapourSynth
1 BVFI
arXiv		 35.43 {4}
arXiv		 Adobe240 - - -
2 BiT++
CVPR		 34.97 {3}
CVPR		 Adobe240 GitHub Stars Request -
3 DeMFI-Netrb(5,3)
ECCV		 34.34 {4}
ECCV		 Adobe240 GitHub Stars - -
4 ALANET 33.34dB 1 August 2020 1 GitHub Stars - -
5 PRF4 -Large 33.32dB 2 February 2020 3 GitHub Stars - -

Back to Top Back to the List of Rankings

Adobe240 (5:8) with synthetic motion blur✖️: PSNR😞>=25dB

RK Model PSNR ↑ Originally
announced		 Official
  repository  		 Practical
model		 VapourSynth
1 VIDUE 28.74dB 4 March 2023 4 GitHub Stars - -
2 FLAVR 27.23dB 4 December 2020 5 GitHub Stars - -
3 UTI-VFI 26.69dB 4 December 2020 6 GitHub Stars - -
4 DeMFI 25.71dB 4 November 2021 7 GitHub Stars - -

Back to Top Back to the List of Rankings

Vimeo-90K triplet: LPIPS😍(SqueezeNet)<=0.014

RK Model LPIPS ↓ Originally
announced		 Official
  repository  		 Practical
model		 VapourSynth
1 CDFI w/ adaP/U 0.008 8 March 2021 9 GitHub Stars - -
2 EDSC_s-𝓛F 0.010 9 June 2020 10 GitHub Stars EDSC_s-𝓛F -
3 DRVI 0.013 11 August 2021 11 - - -

Back to Top Back to the List of Rankings

Vimeo-90K triplet: LPIPS😍<=0.018

RK     Model        LPIPS ↓   
{Input fr.}		 Training
dataset		 Official
  repository  		 Practical
model		 VapourSynth
1 EAFI-𝓛ecp
arXiv		 0.012 {2}
arXiv		 Vimeo-90K triplet - EAFI-𝓛ecp -
2 UGFI 𝓛S
CVPR		 0.0126 {2}
CVPR		 Vimeo-90K triplet - UGFI 𝓛S -
3 SoftSplat - 𝓛F
CVPR		 0.013 {2}
CVPR		 Vimeo-90K triplet GitHub Stars SoftSplat - 𝓛F -
4 FILM-𝓛S
ECCV		 0.0132 {2}
CVPR		 Vimeo-90K triplet GitHub Stars FILM-𝓛S -
5 EDSC_s-𝓛F
TPAMI		 0.016 {2}
arXiv		 Vimeo-90K triplet GitHub Stars EDSC_s-𝓛F -
6 CtxSyn - 𝓛F
CVPR		 0.017 {2}
CVPR		 proprietary - CtxSyn - 𝓛F -
7 PerVFI
CVPR		 0.018 {2}
CVPR		 Vimeo-90K triplet GitHub Stars PerVFI -

Back to Top Back to the List of Rankings

Vimeo-90K triplet: PSNR😞>=36dB

RK     Model        PSNR ↑   
{Input fr.}		 Originally
announced
or Training
dataset		 Official
  repository  		 Practical
model		 VapourSynth
1 MA-GCSPA-triplets
CVPR		 36.85 {2}
CVPR		 Vimeo-90K triplet GitHub Stars - -
2 VFIformer + HRFFM
CVPR
ENH:
arXiv		 36.69 {2}
arXiv		 Vimeo-90K triplet GitHub Stars
ENH:
-		 - -
3 LADDER-L
arXiv		 36.65 {2}
arXiv		 Vimeo-90K triplet - - -
4-5 EMA-VFI 36.64dB 12 March 2023 12 GitHub Stars - -
4-5 VFIMamba
arXiv		 36.64 {2}
arXiv		 Vimeo-90K triplet & X-TRAIN GitHub Stars - -
6 IQ-VFI
CVPR		 36.60 {2}
CVPR		 Vimeo-90K triplet - - -
7 DQBC-Aug 36.57dB 13 April 2023 13 GitHub Stars - -
8 TTVFI 36.54dB 14 July 2022 14 GitHub Stars - -
9 AMT-G 36.53dB 15 April 2023 15 GitHub Stars - -
10 AdaFNIO 36.50dB 16 November 2022 16 GitHub Stars - -
11 FGDCN-L 36.46dB 17 November 2022 17 GitHub Stars - -
12 VFIFT
MM		 36.43 {2}
arXiv		 Vimeo-90K triplet - - -
13 UPR-Net LARGE 36.42dB 18 November 2022 18 GitHub Stars - -
14 EAFI-𝓛ecc 36.38dB 19 July 2022 19 - EAFI-𝓛ecp -
15 H-VFI-Large 36.37dB 20 November 2022 20 - - -
16 UGFI 𝓛1
CVPR		 36.34 {2}
CVPR		 Vimeo-90K triplet - UGFI 𝓛S -
17 VFIT-B
CVPR		 36.33 {2}
arXiv		 ? GitHub Stars - -
18 SoftSplat - 𝓛Lap with ensemble 36.28dB 21 March 2020 22 GitHub Stars SoftSplat - 𝓛F -
19 ProBoost-Net (448x256)
TMM		 36.23 {2}
TMM		 ? - - -
20 NCM-Large 36.22dB 23 July 2022 23 - - -
21-22 IFRNet large 36.20dB 24 May 2022 24 GitHub Stars - -
21-22 RAFT-M2M++
CVPR
ENH:
TPAMI		 36.20 {2}
arXiv		 Vimeo-90K triplet GitHub Stars - -
23-24 EBME-H* 36.19dB 25 June 2022 25 GitHub Stars - -
23-24 RIFE-Large
ECCV		 36.19 {2}
ECCV		 Vimeo-90K triplet GitHub Stars Practical-RIFE 4.25 TensorRT
GitHub Stars
TensorRT
GitHub Stars
ncnn
GitHub Stars
25 ABME 36.18dB 26 August 2021 26 GitHub Stars - -
26 HiFI
arXiv		 36.12 {2}
arXiv		 Pretraining: Raw videos
Training: Vimeo-90K triplet & X-TRAIN		 - - -
27 TDPNetnv w/o MRTM
Access		 36.069 {2}
Access		 Vimeo-90K triplet - TDPNet -
28 FILM-𝓛1
ECCV		 36.06 {2}
ECCV		 Vimeo-90K triplet GitHub Stars FILM-𝓛S -

Back to Top Back to the List of Rankings

Vimeo-90K septuplet: LPIPS😍<=0.032

RK Model LPIPS ↓ Originally
announced		 Official
  repository  		 Practical
model		 VapourSynth
1 RIFE 0.0233 27 November 2020 28 GitHub Stars Practical-RIFE 4.25 TensorRT
GitHub Stars
TensorRT
GitHub Stars
ncnn
GitHub Stars
2 IFRNet 0.0274 27 May 2022 24 GitHub Stars - -
3 VFIT-B 0.0304 27 November 2021 29 GitHub Stars - -
4 ABME 0.0309 27 August 2021 26 GitHub Stars - -

Back to Top Back to the List of Rankings

Vimeo-90K septuplet: PSNR😞>=36dB

RK     Model        PSNR ↑   
{Input fr.}		 Originally
announced
or Training
dataset		 Official
  repository  		 Practical
model		 VapourSynth
1 Swin-VFI
arXiv		 38.04 {6}
arXiv		 Vimeo-90K septuplet - - -
2 JNMR 37.19dB 30 June 2022 30 GitHub Stars - -
3 VFIT-B
CVPR		 36.96 {4}
CVPR		 Vimeo-90K septuplet GitHub Stars - -
4 VRT
arXiv		 36.53 {4}
arXiv		 Vimeo-90K septuplet GitHub Stars - -
5 ST-MFNet 36.507dB 31 November 2021 32 GitHub Stars - -
6 EDENVFI PVT(15,15) 36.387dB 31 July 2023 31 - - -
7 IFRNet
CVPR		 36.37 {2}
CVPR		 Vimeo-90K septuplet GitHub Stars - -
8 RN-VFI
CVPR		 36.33 {4}
CVPR		 Vimeo-90K septuplet - - -
9 FLAVR
WACV		 36.3 {4}
WACV		 Vimeo-90K septuplet GitHub Stars - -
10 DBVI 36.17dB 33 October 2022 33 GitHub Stars - -
11 EDC 36.14dB 30 February 2022 34 GitHub Stars - -

Back to Top Back to the List of Rankings

Appendix 2: Metrics selection for the rankings

Currently, the most commonly used metrics in the existing works on video frame interpolation and video deblurring are: PSNR, SSIM and LPIPS. Exactly in that order.

The main purpose of creating my rankings is to look for the best perceptually-oriented model for practical applications - hence the primary metric in my rankings will be the most common perceptual image quality metric in scientific papers: LPIPS.

At the time of writing these words, in October 2023, in relation to VFI, I have only found another perceptual image quality metric - DISTS in one paper: Access and also in one paper I found a bespoke VFI metric - FloLPIPS [arXiv]. Unfortunately, both of these papers omit to evaluate the best performing models based on the LPIPS metric. If, in the future, some researcher will evaluate LPIPS top-performing models using alternative, better perceptual metrics, I would of course be happy to add rankings based on those metrics.

I would like to use only one metric - LPIPS. Unfortunately still many of the best VFI and video deblurring methods are only evaluated using PSNR or PSNR and SSIM. For this reason, I will additionally present rankings based on PSNR, which will show the models that can, after perceptually-oriented training, be the best for practical applications, as well as providing a source of knowledge for building even better practical models in the future.

I have decided to completely abandon rankings based on the SSIM metric. Below are the main reasons for this decision, ranked from the most important to the less important.

  • The main reason is the following quote, which I found in a paper by researchers at Adobe Research: 21. In the quote they refer to a paper by researchers at NVIDIA: [arXiv].

    We limit the evaluation herein to the PSNR metric since SSIM [57] is subject to unexpected and unintuitive results [39].

  • The second reason is, more and more papers are appearing where PSNR scores are given, but without SSIM: 31 and Access A model from such a paper appearing only in the PSNR-based ranking and at the same time not appearing in the SSIM-based ranking may give the misleading impression that the SSIM score is so poor that it does not exceed the ranking eligibility threshold, while there is simply no SSIM score in a paper.

  • The third reason is, that often the SSIM scores of individual models are very close to each other or identical. This is the case in the SNU-FILM Easy test, as shown in Table 3: [CVPR 2023], where as many as 6 models achieve the same score of 0.991 and as many as 5 models achieve the same score of 0.990. In the same test, PSNR makes it easier to determine the order of the ranking, with the same number of significant digits.

  • The fourth reason is that PSNR-based rankings are only ancillary when a model does not have an LPIPS score. For this reason, SSIM rankings do not add value to my repository and only reduce its readability.

  • The fifth reason is that I want to encourage researchers who want to use only two metrics in their paper to use LPIPS and PSNR instead of PSNR and SSIM.

  • The sixth reason is that the time saved by dropping the SSIM-based rankings will allow me to add new rankings based on other test data, which will be more useful and valuable.

Back to Top Back to the List of Rankings

Appendix 3: List of all research papers from the above rankings

📝 Note: Temporarily, the following list contains full descriptions of those methods that have been removed from the footnotes or not included in the footnotes at all due to the new layout of the tables.

Method Paper     Venue    
ABME
AdaFNIO
ALANET
AMT
BIN Blurry Video Frame Interpolation CVPR
BiT Blur Interpolation Transformer for Real-World Motion from Blur CVPR
BVFI Three-Stage Cascade Framework for Blurry Video Frame Interpolation arXiv
CDFI
CtxSyn Context-aware Synthesis for Video Frame Interpolation CVPR
DBVI
DeMFI DeMFI: Deep Joint Deblurring and Multi-Frame Interpolation with Flow-Guided Attentive Correlation and Recursive Boosting ECCV
DQBC
DRVI
EAFI Error-Aware Spatial Ensembles for Video Frame Interpolation arXiv
EBME
EDC Enhancing Deformable Convolution based Video Frame Interpolation with Coarse-to-fine 3D CNN ICIP
EDENVFI
EDSC Multiple Video Frame Interpolation via Enhanced Deformable Separable Convolution TPAMI
EMA-VFI
FGDCN
FILM FILM: Frame Interpolation for Large Motion ECCV
FLAVR FLAVR: Flow-Agnostic Video Representations for Fast Frame Interpolation WACV
HiFI High-Resolution Frame Interpolation with Patch-based Cascaded Diffusion arXiv
HRFFM Video Frame Interpolation with Region-Distinguishable Priors from SAM arXiv
H-VFI
IFRNet IFRNet: Intermediate Feature Refine Network for Efficient Frame Interpolation CVPR
IQ-VFI IQ-VFI: Implicit Quadratic Motion Estimation for Video Frame Interpolation CVPR
JNMR
LADDER LADDER: An Efficient Framework for Video Frame Interpolation arXiv
M2M Many-to-many Splatting for Efficient Video Frame Interpolation CVPR
MA-GCSPA Exploring Motion Ambiguity and Alignment for High-Quality Video Frame Interpolation CVPR
NCM
PerVFI Perceptual-Oriented Video Frame Interpolation Via Asymmetric Synergistic Blending CVPR
PRF Video Frame Interpolation and Enhancement via Pyramid Recurrent Framework TIP
ProBoost-Net Progressive Motion Boosting for Video Frame Interpolation TMM
RIFE Real-Time Intermediate Flow Estimation for Video Frame Interpolation ECCV
RN-VFI Range-nullspace Video Frame Interpolation with Focalized Motion Estimation CVPR
SoftSplat Softmax Splatting for Video Frame Interpolation CVPR
SSR Video Frame Interpolation with Many-to-many Splatting and Spatial Selective Refinement TPAMI
ST-MFNet
Swin-VFI Video Frame Interpolation for Polarization via Swin-Transformer arXiv
TDPNet Textural Detail Preservation Network for Video Frame Interpolation Access
TTVFI
UGFI Frame Interpolation Transformer and Uncertainty Guidance CVPR
UPR-Net
UTI-VFI
VFIformer Video Frame Interpolation with Transformer CVPR
VFIFT Video Frame Interpolation with Flow Transformer MM
VFIMamba VFIMamba: Video Frame Interpolation with State Space Models arXiv
VFIT Video Frame Interpolation Transformer CVPR
VIDUE
VRT VRT: A Video Restoration Transformer arXiv

Back to Top Back to the List of Rankings

Footnotes

  1. ALANET: Adaptive Latent Attention Network for Joint Video Deblurring and Interpolation [MM 2020] [arXiv] ↩ ↩2

  2. Video Frame Interpolation and Enhancement via Pyramid Recurrent Framework [TIP 2020] ↩

  3. Blurry Video Frame Interpolation [CVPR 2020] [arXiv] ↩

  4. Joint Video Multi-Frame Interpolation and Deblurring under Unknown Exposure Time [CVPR 2023] [arXiv] ↩ ↩2 ↩3 ↩4 ↩5

  5. FLAVR: Flow-Agnostic Video Representations for Fast Frame Interpolation [WACV 2023] [arXiv] ↩

  6. Video Frame Interpolation without Temporal Priors [NeurIPS 2020] [arXiv] ↩

  7. DeMFI: Deep Joint Deblurring and Multi-Frame Interpolation with Flow-Guided Attentive Correlation and Recursive Boosting [ECCV 2022] [arXiv] ↩

  8. AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling [TIP 2022] [arXiv] ↩

  9. CDFI: Compression-Driven Network Design for Frame Interpolation [CVPR 2021] [arXiv] ↩ ↩2

  10. Multiple Video Frame Interpolation via Enhanced Deformable Separable Convolution [TPAMI 2021] [arXiv] ↩

  11. DRVI: Dual Refinement for Video Interpolation [Access 2021] ↩ ↩2

  12. Extracting Motion and Appearance via Inter-Frame Attention for Efficient Video Frame Interpolation [CVPR 2023] [arXiv] ↩ ↩2

  13. Video Frame Interpolation with Densely Queried Bilateral Correlation [IJCAI 2023] [arXiv] ↩ ↩2

  14. TTVFI: Learning Trajectory-Aware Transformer for Video Frame Interpolation [TIP 2023] [arXiv] ↩ ↩2

  15. AMT: All-Pairs Multi-Field Transforms for Efficient Frame Interpolation [CVPR 2023] [arXiv] ↩ ↩2

  16. AdaFNIO: Adaptive Fourier Neural Interpolation Operator for video frame interpolation [arXiv] ↩ ↩2

  17. Flow Guidance Deformable Compensation Network for Video Frame Interpolation [TMM 2023] [arXiv] ↩ ↩2

  18. A Unified Pyramid Recurrent Network for Video Frame Interpolation [CVPR 2023] [arXiv] ↩ ↩2

  19. Error-Aware Spatial Ensembles for Video Frame Interpolation [arXiv] ↩ ↩2

  20. H-VFI: Hierarchical Frame Interpolation for Videos with Large Motions [arXiv] ↩ ↩2

  21. Revisiting Adaptive Convolutions for Video Frame Interpolation [WACV 2021] [arXiv] ↩ ↩2

  22. Softmax Splatting for Video Frame Interpolation [CVPR 2020] [arXiv] ↩

  23. Neighbor Correspondence Matching for Flow-based Video Frame Synthesis [MM 2022] [arXiv] ↩ ↩2

  24. IFRNet: Intermediate Feature Refine Network for Efficient Frame Interpolation [CVPR 2022] [arXiv] ↩ ↩2 ↩3

  25. Enhanced Bi-directional Motion Estimation for Video Frame Interpolation [WACV 2023] [arXiv] ↩ ↩2

  26. Asymmetric Bilateral Motion Estimation for Video Frame Interpolation [ICCV 2021] [arXiv] ↩ ↩2 ↩3

  27. Exploring Discontinuity for Video Frame Interpolation [CVPR 2023] [arXiv] ↩ ↩2 ↩3 ↩4

  28. Real-Time Intermediate Flow Estimation for Video Frame Interpolation [ECCV 2022] [arXiv] ↩

  29. Video Frame Interpolation Transformer [CVPR 2022] [arXiv] ↩

  30. JNMR: Joint Non-linear Motion Regression for Video Frame Interpolation [TIP 2023] [arXiv] ↩ ↩2 ↩3

  31. Efficient Convolution and Transformer-Based Network for Video Frame Interpolation [ICIP 2023] [arXiv] ↩ ↩2 ↩3 ↩4

  32. ST-MFNet: A Spatio-Temporal Multi-Flow Network for Frame Interpolation [CVPR 2022] [arXiv] ↩

  33. Deep Bayesian Video Frame Interpolation [ECCV 2022] ↩ ↩2

  34. Enhancing Deformable Convolution based Video Frame Interpolation with Coarse-to-fine 3D CNN [ICIP 2022] [arXiv] ↩

About

Rankings include: ABME AdaFNIO ALANET AMT BiT BVFI CDFI CtxSyn DBVI DeMFI DQBC DRVI EAFI EBME EDC EDENVFI EDSC EMA-VFI FGDCN FILM FLAVR HiFI H-VFI IFRNet IQ-VFI JNMR LADDER M2M MA-GCSPA NCM PerVFI PRF ProBoost-Net RIFE RN-VFI SoftSplat SSR ST-MFNet Swin-VFI TDPNet TTVFI UGFI UPR-Net UTI-VFI VFIformer VFIFT VFIMamba VFIT VIDUE VRT

Topics

machine-learning computer-vision deep-learning interpolation artificial-intelligence video-processing motion-blur vapoursynth restoration tensorrt deblurring motion-deblurring ncnn frame-interpolation low-level-vision video-interpolation video-deblurring video-frame-interpolation video-restoration rife

Resources

Readme
Activity

Stars

100 stars

Watchers

9 watching

Forks

0 forks
Report repository