Multiscale structural similarity for image quality assessment
The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; none of the 14 sources its claims cite is checked for support yet.
Abstract
The structural similarity image quality paradigm is based on the assumption that the human visual system is highly adapted for extracting structural information from the scene, and therefore a measure of structural similarity can provide a good approximation to perceived image quality. This paper proposes a multiscale structural similarity method, which supplies more flexibility than previous single-scale methods in incorporating the variations of viewing conditions. We develop an image synthesis method to calibrate the parameters that define the relative importance of different scales. Experimental comparisons demonstrate the effectiveness of the proposed method.
Overview
CoChat Summary
TL;DR
This paper proposes a multi-scale structural similarity (MS-SSIM) image quality method that combines contrast and structure comparisons across scales and luminance comparison at the coarsest scale. On the LIVE JPEG/JPEG2000 database, MS-SSIM outperformed PSNR, the Sarnoff model, and single-scale SSIM variants on all reported evaluation criteria.
Key findings
- MS-SSIM achieved the best reported performance: CC 0.969, ROCC 0.966, MAE 3.86, RMS 4.91, and OR 1.16%, compared with the best single-scale SSIM at M=2 (CC 0.963, ROCC 0.959, MAE 4.21, RMS 5.38, OR 2.62%) and PSNR (CC 0.905, ROCC 0.901, MAE 6.53, RMS 8.45, OR 15.7%).
- Single-scale SSIM performance varied with scale; the best single-scale result was at M=2, and single-scale SSIM tended to give higher scores as scale increased.
- The calibrated cross-scale exponents were β1=γ1=0.0448, β2=γ2=0.2856, β3=γ3=0.3001, β4=γ4=0.2363, and α5=β5=γ5=0.1333.
- The authors conclude that multi-scale SSIM provides a meaningful balance between scales and outperforms the best single-scale SSIM and state-of-the-art image quality metrics.
Methods
- The MS-SSIM system iteratively applies a low-pass filter and downsamples by 2; at each scale it computes contrast and structure comparisons, computes luminance comparison only at Scale M, and combines them using exponents normalized so that ∑γj=1.
- Cross-scale parameters were calibrated by image synthesis: 5 scales and 12 distortion levels (range 2^3 to 2^14) were used with 10 original 64×64 images, producing 600 distorted images; 8 subjects selected images of equal perceived quality across scales at a viewing distance of 32 pixels per degree of visual angle.
- Evaluation used the LIVE database with 344 JPEG and JPEG2000 compressed images (bit rate 0.028 to 3.150 bits/pixel), MOS from 13–25 subjective scores, and compared PSNR, Sarnoff JNDmetrix 8.0, single-scale SSIM for M=1–5, and MS-SSIM using CC, MAE, RMS, OR, and ROCC after logistic mapping.
Limitations
- Stated: the image-synthesis calibration approach is “still rather crude,” and the authors are working on a more systematic approach.
- Stated: CSF cannot be directly used to derive the parameters because it is typically measured at visibility threshold with simplified stimuli, whereas the goal is comparing complex structured images at visible distortion levels.
- Evident from the text: calibration used only 10 original 64×64 grayscale images and 8 subjects, including one author, and evaluation was limited to the LIVE JPEG/JPEG2000 database.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 18 Own results and interpretations
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It can also be observed that the single-scale model tends to supply higher scores with the increase of scales.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, for every one of the objective evaluation criteria, multi-scale SSIM model outperforms all the other models, including the best single-scale SSIM model, suggesting a meaningful balance between scales.”
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
“Experiments show that with an appropriate parameter settings, the multi-scale method outperforms the best single-scale SSIM model as well as state-of-the-art image quality metrics.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In general, the subjects agreed with each other on each image more than they agreed with themselves across different images.”
plainstates an associationn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resulting parameters we obtained are β 1 = γ 1 = 0.0448, β 2 = γ 2 = 0.2856, β 3 = γ 3 = 0.3001, β 4 = γ 4 = 0.2363, and α 5 = β 5 = γ 5 = 0.1333, respectively.”
plainn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“However, the SSIM index algorithm introduced in [5] is a single-scale approach.”
What the sources say
- Not matchedcites [5] [5] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, 'Image quality assessment: From error measurement to structural similarity,' IEEE Trans. Image Processing , vol. 13, Jan. 2004. -
More than one paper fits this citation.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
10 more claims
Create a free account to see all 33 claims and every source verdict.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 13
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Further, as research in quality assessment has demonstrated, incorporating multiscale information when assessing quality produces QA algorithms that perform better in terms of correlation with human perception [13], [29].”
From No-Reference Image Quality Assessment in the Spatial Domain (2012) · cites this paper as [13]
Citations
Cites
Image quality assessment: from error visibility to structural similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · IEEE Transactions on Image Processing · doi:10.1109/tip.2003.819861 · Cited by 57,892
Self-citationA universal image quality index
Wang, Z., & Bovik, A. C. (2002) · IEEE Signal Processing Letters · doi:10.1109/97.995823 · Cited by 5,757
Self-citationImage quality measures and their performance
Eskicioglu, A. M., & Fisher, P. S. (1995) · IEEE Transactions on Communications · doi:10.1109/26.477498 · Cited by 2,117 · cited 1×
“Traditional perceptual image quality assessment methods are based on a bottom-up approach which attempts to simulate the functionality of the relevant early human visual system (HVS) components.”
BackgroundA Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients
Portilla, J., & Simoncelli, E. P. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026553619983 · Cited by 1,806 · cited 2×
“These synthesis methods provide a powerful and efficient means of testing a model, and have the added benefit that the resulting images suggest improvements that might be made to the model [11].”
MethodSelf-citation(1995) · Medical Entomology and Zoology · Cited by 1,212
Pyramid-based texture analysis/synthesis
Heeger, D. J., & Bergen, J. R. (1995) · Proceedings of the 22nd annual conference on Computer graphics and interactive techniques - SIGGRAPH '95 · doi:10.1145/218380.218446 · Cited by 1,095
Video quality assessment based on structural distortion measurement
Wang, Z., Lu, L., & Bovik, A. C. (2004) · Signal Processing: Image Communication · doi:10.1016/s0923-5965(03)00076-6 · Cited by 1,024
Self-citationImage Quality Assessment: From Error Measurement to Structural Similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · Cited by 919
Self-citationTeo, P. C., & Heeger, D. J. (2002) · Proceedings of 1st International Conference on Image Processing · doi:10.1109/icip.1994.413502 · Cited by 471
Objective video quality assessment
(2003) · Handbook of Video Databases · doi:10.1201/9780203489864-45 · Cited by 359
Research works printed but not linked: 2
- 0, JNDmetrix Technology, Sarnoff Corporation
- 2000, Final report from the video quality experts group on the validation of objective models of video quality assessment, VQEG
Cited by
Items 1,301–1,400 of 6,047
Stacked Deep Multi-Scale Hierarchical Network for Fast Bokeh Effect Rendering from a Single Image
Dutta, S., Das, S. D., Shah, N. A., & Tiwari, A. K. (2021) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw53098.2021.00272 · Cited by 37
“5, it can be inferred that stage-2 training removes artifacts in the background, and the third and fourth row shows that LMS-SSIM helps in the better reconstruction of the foreground.”
MethodInfluentialDeep Learning-Based Compression for Phase-Only Hologram
Ko, H., & Kim, H.-Y. (2021) · IEEE Access · doi:10.1109/access.2021.3084800 · Cited by 19
“As a performance evaluation measurement, PSNR or (MS)SSIM [42], [43] is commonly used as an objective image quality metric.”
MethodNTIRE 2021 Challenge for Defocus Deblurring Using Dual-pixel Images: Methods and Results
Abuolaim, A., Timofte, R., Brown, M. S., et al. (2021) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw53098.2021.00070 · Cited by 26
“We adopt a loss function which is the weighted sum of MSE loss and SSIM loss.”
MethodInfluentialZhang, X., Li, Z., Zhang, T., Liu, F., Tang, X., Chen, P., & Jiao, L. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3106681 · Cited by 32
“3) Evaluation Metrics: We choose a peak signal-to-noise ratio (PSNR) [50] and Structure SIMilarity (SSIM) [51] as the evaluation metrics for experiment results.”
MethodQuality Assessment of End-to-End Learned Image Compression: The Benchmark and Objective Measure
Li, Y., Wang, S.-Q., Zhang, X.-F., Wang, S.-S., Ma, S.-W., & Wang, Y. (2021) · ACM Multimedia · doi:10.1145/3474085.3475569 · Cited by 36
“Traditional metrics, i.e., PSNR, SSIM, MS-SSIM, VIF, GMSD and NLPD are directly evaluated on the 20% testing data.”
MethodInfluentialTGP-PCQA: Texture and geometry projection based quality assessment for colored point clouds
He, Z., Jiang, G., Yu, M., et al. (2022) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2022.103449 · Cited by 26
Wang, Y., Shi, X., Lei, L., & Fung, J. C. H. (2022) · Monthly Weather Review · doi:10.1175/mwr-d-21-0288.1 · Cited by 29
“(7) of Wang et al. (2003).”
AVQBits—Adaptive Video Quality Model Based on Bitstream Information for Various Video Applications
Rao, R. R. R., Göring, S., & Raake, A. (2022) · IEEE Access · doi:10.1109/access.2022.3195527 · Cited by 29
Learning based Multi-modality Image and Video Compression
Lu, G. S., Zhong, T., Geng, J., Hu, Q., & Xu, D. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00599 · Cited by 43
A holistic survey of multipath wireless video streaming
Afzal, S., Testoni, V., Rothenberg, C. E., Kolan, P., & Bouazizi, I. (2023) · Journal of Network and Computer Applications · doi:10.1016/j.jnca.2022.103581 · Cited by 60
Seeing Through Clouds in Satellite Images
Zhao, M., Olsen, P. A., & Chandra, R. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3239592 · Cited by 55
“3We used a custom implementation of MS-SSIM that appropriately ignores masked out regions of the image.”
MethodInfluentialLecompagnon, J., Hirsch, P. D., Rupprecht, C., et al. (2023) · Scientific Reports · doi:10.1038/s41598-023-30494-2 · Cited by 27
Hyperspectral Image Compression via Cross-Channel Contrastive Learning
Guo, Y., Chong, Y., & Pan, S. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3282186 · Cited by 29
“Meanwhile, the peak signal-to-noise ratio (PSNR), multiscale structural similarity metric (MS-SSIM) [46], and spectral angle mapping (SAM) [47] are used to quantitatively measure the distortion.”
BackgroundMethodHigh-Fidelity and Freely Controllable Talking Head Video Generation
Gao, Y., Zhou, Y., Wang, J., Li, X., Xiang, M., & Lü, Y. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00543 · Cited by 62
“We use L1, MS-SSIM [53,54], and PSNR to evaluate the low-level similarity between the synthesis and the driving images.”
MethodMRI motion artifact reduction using a conditional diffusion probabilistic model (MAR‐CDPM)
Safari, M., Yang, X., Fatemi, A. B., & Archambault, L. (2023) · Medical Physics · doi:10.1002/mp.16844 · Cited by 32
Deep Unrestricted Document Image Rectification
Feng, H., Liu, S., Deng, J., Zhou, W., & Li, H. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3347094 · Cited by 45
“Then, it is computed with a 5-level-pyramid, following original MSSIM [68].”
BackgroundMethodImage quality assessment using edge based features
Attar, A., Shahbahrami, A., & Rad, R. M. (2016) · Multimedia tools and applications · doi:10.1007/s11042-015-2663-9 · Cited by 24
Objective assessment of tone mapping algorithms
Yeganeh, H., & Wang, Z. (2010) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2010.5651778 · Cited by 27
Self-citationTemporal information assisted video quality metric for multimedia
Liu, T.-J., Liu, K.-H., & Liu, H. (2010) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2010.5583094 · Cited by 27
“Here we choose the multi-scale structural similarity (MS-SSIM) index [13] as the basis of our video quality metric because it has very good performance in measuring spatial distortions and also can incorporate image details at different resolutions.”
MethodA new blind image quality framework based on natural color statistic
Wang, Q., Chu, J., Xu, L., & Chen, Q. (2015) · Neurocomputing · doi:10.1016/j.neucom.2015.09.057 · Cited by 29
Subjective and objective evaluation of image inpainting quality
Tiefenbacher, P., Bogischef, V., Merget, D., & Rigoll, G. (2015) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2015.7350838 · Cited by 27
“The popular PSNR and SSIM/MSSSIM measures were not suited to judge inpainting quality.”
MethodInfluentialVisual Image Quality Assessment Technique using FSIM
Kumar, R., & Moyal, V. (2013) · International Journal of Computer Applications Technology and Research · doi:10.7753/ijcatr0203.1008 · Cited by 26
Dynamic adaptive streaming for multi-viewpoint omnidirectional videos
Corbillon, X., Francesca, D. S., Simon, G., & Pascal Frossard, B. (2018) · ACM Multimedia Systems Conference (MMSys) · doi:10.1145/3204949.3204968 · Cited by 28
“Figure 5a represents the median MS-SSIM [39] inside the video displayed in the viewports for various available bandwidth.”
BackgroundLi, J., Krasula, L., Baveye, Y., et al. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2019.2903722 · Cited by 27
“OBJECTIVE VIDEO QUALITY METRICS Generally, the state-of-the-art objective video quality metrics such as PSNR, SSIM [38], MS-SSIM [39], MOVIE [40], VQM [41], VQM-VFD [42], SSIMplus [43], VMAF [44] are used to predict the exact perceptual quality of the video sequences, and do not consider the threshold for acceptabilit…”
MethodFast Blind Quality Assessment of DIBR-Synthesized Video Based on High-High Wavelet Subband
Sandic-Stankovic, D. D., Kukolj, D., & Callet, P. L. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2919416 · Cited by 28
“Scatter plots of subjective scores versus the proposed metric NR_MWT_lienergy and other tested metrics: Q [43], APT [6], NIQSV+ [40], NIQSV [39], SVQM [33], MP-PSNR [9], MW-PSNR_red [7], PSNR, SSIM, MS-SSIM [53], ADD-SSIM [56].”
BackgroundLearning to Predict Streaming Video QoE: Distortions, Rebuffering and MemoryPreprint
Bampis, C. G., & Bovik, A. C. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1703.00633 · Cited by 52
“These include standard frame-based techniques (FR-IQA) such as SSIM [11], [12] and MS-SSIM [13], temporal FR-VQA methods such as VQM VFD [14], MOVIE [15], ST-MAD [16], VMAF [17] and FLOSIM [18] and reduced-reference models like STRRED [19].”
InfluentialSelf-citationImage quality assessment based on Prewitt magnitude
Zhang, H., Zhu, Q., Fan, C., & Deng, D. (2013) · AEU - International Journal of Electronics and Communications · doi:10.1016/j.aeue.2013.04.001 · Cited by 27
Xu, Y., Wei, M., & Kamruzzaman, M. M. (2021) · Future Generation Computer Systems · doi:10.1016/j.future.2020.11.015 · Cited by 39
Maruyama, S. (2023) · Physical and Engineering Sciences in Medicine · doi:10.1007/s13246-023-01280-1 · Cited by 29
HDIQA: A Hyper Debiasing Framework for Full Reference Image Quality Assessment
Zhou, M., Wang, H., Wei, X., Feng, Y., Luo, J. T., Pu, H., Zhao, J., Wang, L., Chu, Z., Wang, X., et al. (2024) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2024.3353573 · Cited by 30
“The methodologies employed for comparison include manual-crafted feature-based methods [8], [9], [11], [41], [42] and Authorized licensed use limited to the terms of the applicable license agreement with IEEE.”
MethodNo-Reference Point Cloud Quality Assessment Through Structure Sampling and Clustering Based on Graph
Wu, X., He, Z., Jiang, G., et al. (2024) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2024.3482173 · Cited by 27
“In addition to the methods executed in 3D space, PCs can be simplified by transforming them into regular images via 2D projection, followed using 2D IQA methods like SSIM [37], MS-SSIM [45], IW-SSIM [46], and VIFP [47] to predict PC quality.”
Structural-Equation-Modeling-Based Indicator Systems for Image Quality Assessment
Wang, C., Lin, J., & Li, X. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3567814 · Cited by 31
Zhang, C., Pan, J. S., Chen, S., Wang, T., & Sun, D. (2015) · Neurocomputing · doi:10.1016/j.neucom.2015.01.105 · Cited by 27
Consistent Visual Quality Control in Video Coding
Xu, L., Li, S., Ngan, K. N., & Ma, L. (2013) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2013.2243657 · Cited by 27
Spatial-temporal Video Quality Metric based on an estimation of QoE
Amirshahi, S. A., & Larabi, M. (2011) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex.2011.6065718 · Cited by 28
“html [19] Wang, Z.”
Entropy of Primitive: From Sparse Representation to Visual Information Evaluation
Ma, S., Zhang, X., Wang, S., Zhang, J., Sun, H., & Gao, W. (2015) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2015.2511838 · Cited by 26
“Therefore, the SSIM and its variants [15]–[17] have been well advocated in both academic and industry endeavors due to the best tradeoff between accuracy and efficiency.”
BackgroundPerceptual quality assessment of 3D synthesized images
Farid, M. S., Lucenteforte, M., & Grangetto, M. (2017) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2017.8019307 · Cited by 25
“SSIM [33], MSSIM [34], PSNR, and VSNR [35] for performance comparison.”
MethodStatistical Evaluation of Visual Quality Metrics for Image Denoising
Egiazarian, K., Ponomarenko, N. N., Лукин, В. В., & Єремеєв, О. (2018) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2018.8462294 · Cited by 31
Contrast and Visual Saliency Similarity-Induced Index for Assessing Image Quality
Jia, H.-Z., Zhang, L., & Wang, T.-H. (2017) · IEEE Access · doi:10.1109/access.2018.2878739 · Cited by 25
“The proposed method was compared with state-of-the-art methods, such as the SSIM [5], MS-SSIM [6], IW-SSIM [8], VIF [10], MAD [11], FSIM [14], GMSD [16], VSI [15] and MCSD [17].”
BackgroundMethodInfluentialLearned Scalable Image Compression with Bidirectional Context Disentanglement Network
Zhang, Z., Chen, Z., Lin, J., & Li, W. (2019) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2019.00249 · Cited by 28
“Here, we take L1 norm and MS-SSIM (proposed in [21]) as the mentioned distance functions to train our model in this paper.”
MethodDeep Image Compression with Residual Learning
Li, W., Sun, W., Zhao, Y., et al. (2020) · Applied Sciences · doi:10.3390/app10114023 · Cited by 26
“Distortion is often assessed by the distance between x and ˜ x , D ( x , ˜ x ), namely PSNR (Peak Signal to Noise Ratio), or the perceptual measure such as MS-SSIM (Multi-scale Structural Similarity) [13].”
BackgroundUGC-VIDEO: Perceptual Quality Assessment of User-Generated Videos
Li, Y., Meng, S., Zhang, X., Wang, S., Wang, Y., & Ma, S. (2020) · IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) · doi:10.1109/mipr49039.2020.00015 · Cited by 41
3D Scene Compression through Entropy Penalized Neural Representation Functions
Bird, T., Ballé, J., Singh, S., & Chou, P. A. (2021) · Picture Coding Symposium · doi:10.1109/pcs50896.2021.9477505 · Cited by 37
“Although cNeRF is trained to minimize the squared error between renderings and the true images (and therefore maximize PSNR), cNeRF also outperforms HEVC + LLFF in both MS-SSIM (Wang et al., 2003) and LPIPS (Zhang et al., 2018).”
Data-Driven Cloud Clustering via a Rotationally Invariant Autoencoder
Kurihana, T., Moyer, E. J., Willett, R. M., Gilton, D., & Foster, I. T. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3098008 · Cited by 32
“The fourth term is the multiscale structure similarity index (MS-SSIM) [72], a multiband version of the SSIM [73] index often used in computer vision to assess image similarity.”
MethodInfluentialReal‐time Monte Carlo Denoising with Weight Sharing Kernel Prediction Network
Fan, H., Wang, R., Huo, Y., & Bao, H. (2021) · Computer Graphics Forum · doi:10.1111/cgf.14338 · Cited by 47
Bai, Y., Liu, X., Zuo, W., Wang, Y., & Ji, X. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01177 · Cited by 39
“Most recent methods for lossy image compression adopt variational auto-encoder (VAE) architecture [17, 18] based on transform coding [11], where the rate is modeled by the entropy of quantized latent variables and the reconstruction distortion is measured by PSNR or MS-SSIM.”
MethodA gradient-based framework for 3D print appearance optimization
Nindel, T. K., Iser, T., Rittig, T., Wilkie, A., & Křivánek, J. (2021) · ACM Transactions on Graphics · doi:10.1145/3450626.3459844 · Cited by 30
Geo-Context Aware Study of Vision-Based Autonomous Driving Models and Spatial Video Data
Jamonnak, S., Zhao, Y., Huang, X., & Amiruzzaman, M. (2021) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2021.3114853 · Cited by 27
“Here, we implement an image comparison algorithm based on Structural Similarity Index Measure (SSIM), which automatically finds the most different images [54].”
MethodFu, B., Chen, C., Henniger, O., & Damer, N. (2022) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv51458.2022.00119 · Cited by 35
“Full-reference IQA algorithms require an original image for comparison, e.g., peak signal-to-noise ratio (PSNR) [24], and structural similarity index measure (SSIM) [52].”
BackgroundVisual Quality Assessment of Stereoscopic Image and Video: Challenges, Advances, and Future Trends
Su, C.-C., Moorthy, A. K., & Bovik, A. C. (2014) · Springer eBooks · doi:10.1007/978-3-319-10368-6_7 · Cited by 33
Self-citationHuang, H., Chen, Q., Lin, L., Cai, M., Zhang, Q., Iwamoto, Y., Han, X., Furukawa, A., Kanasaki, S., Chen, Y., et al. (2022) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2022.3153406 · Cited by 30
“MS-SSIM [16], [17] is originally proposed for image quality assessment, where the multi-scale structural information is captured from local neighborhood, which is detailly described in Section-III-D.”
BackgroundWang, Y., Chai, X., Gan, Z., Zhang, Y., Chen, X., & He, X. (2022) · Applied Intelligence · doi:10.1007/s10489-022-03597-y · Cited by 27
FVC: An End-to-End Framework towards Deep Video Compression in Feature Space
Hu, Z., Xu, D., Lu, G. S., Jiang, W., Wang, W., & Liu, S. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3210652 · Cited by 43
“PSNR and MS-SSIM [39] are the two most common metrics for evaluating the compression distortion.”
MethodZhu, Z., Ping, W., Qian, Z., Li, S., & Zhang, X. (2022) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2022.3224243 · Cited by 26
“In addition to the removal rate, we also paid attention to the quality of the processed images, therefore, we used PSNR and MS-SSIM [45] as criteria to evaluate the proposed image.”
MethodQ-LIC: Quantizing Learned Image Compression With Channel Splitting
Sun, H., Yu, L., & Katto, J. (2022) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2022.3231789 · Cited by 40
“Besides, the dynamic range of various channels are different, so channel-wise rather than layer-wise quantization is adopted.”
Kim, G., Shim, H., Kim, H., Choi, Y., Kim, J., & Yang, E. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00590 · Cited by 59
“To compare this ability with baselines quantitatively, we measure frequently used metrics for reconstruction - SSIM [35] , Multi-scaled (MS) SSIM [36], LPIPS [41] and MSE - on randomly selected 20 videos in the testset of VoxCeleb1.”
MethodEnd-to-End Learning-Based Image Compression With a Decoupled Framework
Zhang, Z., Esenlik, S., Wu, Y., Wang, M., Zhang, K., & Zhang, L. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3313974 · Cited by 34
“…to the common training and test conditions (CTTC) of JPEG AI [21], the objec-tive quality evaluation is performed by computing seven quality metrics, including MS-SSIM [22], IW-SSIM [23], VMAF [24], VIF [25], PSNR-HVS [26], NLPD [27] and FSIM [28], between the compressed and original image.”
MethodPerceptual quality assessment of 3D dynamic meshes: Subjective and objective studies
Torkhani, F., Wang, K., & Chassery, J. (2015) · Signal Processing Image Communication · doi:10.1016/j.image.2014.12.008 · Cited by 36
“• Two full-reference, 2D still-image perceptual quality metrics: the Multi-Scale Structural SIMilarity index (MS S S IM) [42] and the Visual Information Fidelity (VIF) measure [43].”
Hybrid Distortion Ranking Tuned Bitstream-Layer Video Quality Assessment
Chen, Z., Liao, N., Gu, X., Wu, F., & Shi, G. (2015) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2015.2441432 · Cited by 26
“144) [4], structural similarity (SSIM) [5], visual information fidelity (VIF) [6], multiscale SSIM [7], and peak signalto-noise ratio (PSNR)] in terms of three types of video artifacts (coding, slicing, and freezing) in the literature.”
BackgroundMethodInfluentialTowards Unsupervised Single-channel Blind Source Separation Using Adversarial Pair Unmix-and-remix
Hoshen, Y. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682375 · Cited by 39
Digital Halftone Database (DHD): A Comprehensive Analysis on Halftone Types
Guo, J., & Sankarasrinivasan, S. (2018) · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) · doi:10.23919/apsipa.2018.8659732 · Cited by 26
“Under the quality assessment, the widely adopted structural similarity index [17] is used for the study.”
MethodA robust background initialization algorithm with superpixel motion detection
Xu, Z., Min, B., & Cheung, R. C. C. (2018) · Signal Processing Image Communication · doi:10.1016/j.image.2018.07.004 · Cited by 26
Deep Video Deblurring: The Devil is in the Details
Gast, J., & Roth, S. (2019) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2019.00475 · Cited by 28
“clude DBN 192;1, a single-image variant of our baseline. The detailed results are shown in Table3. Interest7 Table 3. Deblurring performance on the GOPRO dataset of [34] reported as PSNR [24] / MSSIM [53]. Method #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 avg Reference Input 28.77/.938 27.76/.941 26.58/.881 29.83/.976 26.50/.…”
InfluentialUnsupervised Adversarial Correction of Rigid MR Motion Artifacts
Armanious, K., Tanwar, A., Abdulatif, S., Küstner, T., Gatidis, S., & Yang, B. (2020) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi45749.2020.9098570 · Cited by 31
“The multi-scale structural similarity index (MS-SSIM) has been widely used in the literature as a metric to evaluate the performance of deep learning models [24].”
MethodNo-Reference Quality Evaluation of Stereoscopic Video Based on Spatio-Temporal Texture
Yang, J., Zhao, Y. S., Jiang, B., Lu, W. B., & Gao, X. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2019.2961209 · Cited by 26
“SSIM and MS-SSIM estimate the quality by comparing the luminance, contrast and structure of the reference image and the distorted image.”
MethodInfluentialProjected Distribution Loss for Image Enhancement
Delbracio, M., Talebi, H., & Milanfar, P. (2021) · IEEE International Conference on Computational Photography (ICCP) · doi:10.1109/iccp51581.2021.9466271 · Cited by 46
“Note that in contrast to PSNR and MS-SSIM, lower LPIPS means higher similarity to the ground truth.”
MethodInfluentialFast structural similarity index algorithm
Chen, M., & Bovik, A. C. (2010) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2010.5495310 · Cited by 30
Self-citationJohnston, N., Vincent, D., Minnen, D. C., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1703.10114 · Cited by 26
“This dB transform is also supported by the original MS-SSIM [27] publication, which showed the mean opinion score is linearly correlated with the MS-SSIM score after transforming that score to the log domain.”
BackgroundMethodInfluentialDeepObfuscator: Adversarial Training Framework for Privacy-Preserving Image ClassificationPreprint
Li, A., Guo, J., Yang, H., & Chen, Y. (2019) · arXiv (Cornell University) · Cited by 31
G-VAE: A Continuously Variable Rate Deep Image Compression FrameworkPreprint
Cui, Z., Wang, J., Bai, B. B., Guo, T., & Feng, Y. (2020) · arXiv (Cornell University) · Cited by 29
UNet 3+: A Full-Scale Connected UNet for Medical Image SegmentationPreprint
Huang, H., Lin, L., Tong, R., Hu, H., Zhang, Q., Iwamoto, Y., Han, X., Chen, Y., & Wu, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.08790 · Cited by 135
“The MS-SSIM loss function of p and g is defined as: (5) where 𝑀 is the total number of the scales, 𝜇 c , 𝜇 d and 𝜎 c , 𝜎 c are the mean and standard deviations of p and g, 𝜎 cd denotes their covariance.”
MethodInfluentialConditional Diffusion Models for Semantic 3D Medical Image SynthesisPreprint
Dorjsembe, Z., Pao, H.-K., Odonchimed, S., & Xiao, F. (2023) · doi:10.36227/techrxiv.23723787.v2 · Cited by 33
Accelerated Cine Cardiac MRI Using Deep Learning‐Based Reconstruction: A Systematic Evaluation
Pednekar, A. S., Kocaoğlu, M., Wang, H., Tanimoto, A. A., Tkach, J. A., Lang, S. M., & Taylor, M. D. (2023) · Journal of Magnetic Resonance Imaging · doi:10.1002/jmri.29081 · Cited by 26
“The training of the Adaptive-CS-Net involved a loss function that was de fi ned as a weighted sum of multiscale structural similarity index measure (MSSIM), 27 aiming to preserve structural and contrast information while preserving luminance through the L1 norm.”
A direct sampling-based deep learning approach for inverse medium scattering problems
Ning, J., Han, F., & Zou, J. (2023) · Inverse Problems · doi:10.1088/1361-6420/ad0dba · Cited by 25
“The TV refers to the total variation which is commonly used as the regularizer in image reconstruction, and SSIM [49] is the structural similarity function which is a common metric for measuring the perceptual distance between two images and was also added to the loss function in [21]. α 1 , α 2…”
MethodShen, W., Zhou, M., Chen, Y., Wei, X.-K., Feng, Y., Pu, H.-Y., & Jia, W.-J. (2024) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr52734.2025.01676 · Cited by 53
“…and state-of-the-art image quality assessment techniques, including both feature-based and deep learning approaches, including the PSNR, SSIM [33], MS-SSIM [32], VIF [26], MAD [16], FSIM [37], VSI [38], GMSD [34], NLPD [15], PieAPP [23], LPIPS [39], DISTS [4], DeepWSD [17], and TOPIQ-FR [2].”
InfluentialLearning Efficient Data Representations With Orthogonal Sparse Coding
Schütze, H., Barth, E., & Martinetz, T. (2016) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2016.2557065 · Cited by 25
Current Trends and Advances in Image Quality Assessment
Okarma, K. (2019) · Elektronika ir Elektrotechnika · doi:10.5755/j01.eie.25.3.23681 · Cited by 26
Channel-Level Variable Quantization Network for Deep Image Compression
Zhong, Z., Akutsu, H., & Aizawa, K. (2020) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2020/65 · Cited by 33
“Evaluating on the Kodak dataset, this toy model yields an average MSSSIM [Wang et al., 2003] of 0.922 at an average rate of 0.112 bits per pixel (BPP).”
MethodInfluentialDeep Multi-Scale Features Learning for Distorted Image Quality Assessment
Zhou, W., & Chen, Z. (2021) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas51556.2021.9401285 · Cited by 36
“The proposed method is compared with ten IQA models, i.e., PSNR, SSIM [10], MS-SSIM [11], IW-SSIM [12], FSIMc [13], VIF [14], SPSIM [15], LLM [22], DeepSim [23], and DIQaM-FR [24].”
MethodInfluentialPanoramic Image Reflection Removal
Hong, Y., Zheng, Q., Zhao, L., Jiang, X., Kot, A. C., & Shi, B. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.00767 · Cited by 26
“Following the evaluation for existing reflection removal meth-ods [35, 40], we utilize PSNR [8], SSIM [39], NCC [47], and LMSE [4] as error metrics.”
BackgroundMethodInfluentialAmirpour, H., Menon, V. V., Afzal, S., et al. (2022) · ACM Multimedia Systems Conference (MMSys) · doi:10.1145/3524273.3532892 · Cited by 25
“The quality of encodings is evaluated and provided using the following objective quality metrics: VMAF, MS-SSIM, SSIM, and PSNR. Finally, the encoding run-time complexity of the sequences is evaluated using standard Amazon Web Services (AWS) instances, namely c5.xlarge, c5.”
MethodInfluentialHMFVC: A Human–Machine Friendly Video Compression Scheme
Huang, Z., Jia, C., Wang, S., et al. (2022) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2022.3207596 · Cited by 25
“A large number of video compression methods are proposed to achieve high-quality reconstruction under signal fidelity inspired metrics (e.g., PSNR, SSIM [20], etc .) or perceptual metrics (e.g., LPIPS [21], JND [22], etc .)”
Differentiable Rendering Using RGBXY Derivatives and Optimal Transport
Xing, J., Luan, F., Yan, L., Hu, X., Qian, H., & Xu, K. (2022) · ACM Transactions on Graphics · doi:10.1145/3550454.3555479 · Cited by 37
“We provide an evaluation with both 3D and 2D metrics, i.e., a 3D error metric MAE which measures the mean absolute error between vertices of the optimized and target meshes, and 2D error metrics including PSNR, SSIM [Wang et al. 2003], RMSE and a perceptual score LPIPS [Zhang et al. 2018].”
MethodRobustness Analysis of Deep Learning-Based Lung Cancer Classification Using Explainable Methods
Malafaia, M., Silva, F., Neves, I., Pereira, T., & Oliveira, H. P. (2022) · IEEE Access · doi:10.1109/access.2022.3214824 · Cited by 25
“…more classical approaches, such as the Root Mean Squared Error (RMSE), as well as techniques that try to approximate the perceptual quality assessment of the Human Visual System (HVS), such as the Structural SIMilarity (SSIM) index and the Multi-Scale Structural SIMilarity (MS-SSIM) index [34].”
MethodBidirectional Translation Between UHD-HDR and HD-SDR Videos
Yao, M., He, D., Li, X. S., Pan, Z., & Xiong, Z. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3239656 · Cited by 26
“We select five quantitative metrics for fidelity evaluation, including PSNR, mPSNR[57],SSIM,MS-SSIM[58]andHDR-VDP(2.2.1)[59].”
MethodInfluentialCLONeR: Camera-Lidar Fusion for Occupancy Grid-Aided Neural Representations
Carlson, A., Ramanagopal, M. S., Tseng, N., et al. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3262139 · Cited by 25
“CLONeR meets or exceeds DSNeRF* with respect to PSNR and MSSIM while scoring marginally lower with respect to LPIPS.”
MethodPerceptual Quality Assessment for Digital Human Heads
Zhang, Z., Zhou, Y., Sun, W., Min, X., Wu, Y., & Zhai, G. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10095347 · Cited by 37
“The performance results are shown in Table 1, from which we can several inspections: (a) The proposed method achieves the higher performance among all the competitors and is about 0.03 ahead of the second place competitor VIF (0.9286 vs. 0.8898) in terms of SRCC, which shows the effectiveness of the proposed method fo…”
MethodInfluentialDeep learning-based hair removal for improved diagnostics of skin diseases
El‐Shafai, W., El‐Fattah, I. A., & Taha, T. E. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-16646-6 · Cited by 31
“The HRM model is evaluated using four key parameters: Structural Similarity Index (SSIM) [45], Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR) [46], and Multi-Scale Structural Similarity (MS-SSIM) [47].”
A Full-Reference Image Quality Assessment Method via Deep Meta-Learning and Conformer
Lang, S., Liu, X., Zhou, M., Luo, J. T., Pu, H., Zhuang, X., Wang, J., Wei, X., Zhang, T., Feng, Y., et al. (2023) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2023.3308349 · Cited by 28
“To verify the validity of our proposed model, 13 typical methods were selected for comparison: 9 traditional FR-IQA methods—PSNR, SSIM, MS-SSIM [39], VSI, MAD [36], VIF, NLPD [40], GMSD [41], and 4 methods based on deep learning—WaDIQaM [42], PieAPP [43], LPIPS and DISTS.”
MethodOptimizing Multiscale SSIM for Compression via MLDS
Charrier, C., Knoblauch, K., Maloney, L. T., Bovik, A. C., & Moorthy, A. K. (2012) · IEEE Transactions on Image Processing · doi:10.1109/tip.2012.2210723 · Cited by 25
Self-citationOkarma, K. (2013) · Elektronika ir Elektrotechnika · doi:10.5755/j01.eee.19.10.5908 · Cited by 21
Improving chemical species tomography of turbulent flows using covariance estimation
Grauer, S. J., Hadwin, P. J., & Daun, K. J. (2017) · Applied Optics · doi:10.1364/ao.56.003900 · Cited by 25
Video quality assessment by compact representation of energy in 3D-DCT domain
He, L., Lu, W. B., Jia, C., & Hao, L. (2017) · Neurocomputing · doi:10.1016/j.neucom.2016.08.143 · Cited by 26
Helmrich, C. R., Siekmann, M., Becker, S., Bosse, S., Marpe, D., & Wiegand, T. (2020) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp40776.2020.9054089 · Cited by 34
“This (cid:35)(cid:36) approach is not new and has been applied in a number of VQA methods, most prominently the MS-SSIM [3].”
MethodAlgarni, A. D., El‐Banby, G. M., Soliman, N. F., El‐Samie, F. E. A., & Iliyasu, A. M. (2020) · Electronics · doi:10.3390/electronics9061046 · Cited by 26
“Intuitionistic Fuzzy Sets Fuzzy processing is a common component employed in the pre-processing step of many image processing techniques [10].”
BackgroundMethodQuantifying Visual Image Quality: A Bayesian View
Duanmu, Z., Liu, W., Wang, Z., & Wang, Z. (2021) · Annual Review of Vision Science · doi:10.1146/annurev-vision-100419-120301 · Cited by 38
“Most of them apply the SSIM index in the sub-band at different spatial locations [108], orientations [80, 135], and frequency content [97, 122, 129] to simulate the characteristics of primary visual cortex.”
MethodSelf-citationInfrared and visible image fusion based on NSST and phase consistency adaptive DUAL channel PCNN
Xie, Q., Ma, L., Guo, Z., et al. (2023) · Infrared Physics & Technology · doi:10.1016/j.infrared.2023.104659 · Cited by 24
Deep learning-assisted medical image compression challenges and opportunities: systematic review
Bourai, N. E. H., Merouani, H. F., & Djebbar, A. (2024) · Neural Computing and Applications · doi:10.1007/s00521-024-09660-8 · Cited by 35
“Multi-scale structural similarity index (MS-SSIM) and complex wavelet structural similarity (CW-SSIM) An extension of SSIM, known as Multi-Scale SSIM (MS-SSIM), incorporates multiple scales [298].”
A new image quality metric based on human visual system
Dimauro, G. (2012) · Proceedings of the ... IEEE International Conference on Virtual Environments, Human-Computer Interfaces and Measurement Systems./Proceedings of the ... IEEE International Conference on Virtual Environments, Human-Computer Interfaces and Measurement Systems · doi:10.1109/vecims.2012.6273216 · Cited by 24
“The metrics MSSIM [7] and UQI [8] are structural similarity based metrics.”
MethodPerSIM: Multi-resolution image quality assessment in the perceptually uniform color domain
Temel, D., & AlRegib, G. (2015) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2015.7351087 · Cited by 27
“distorted images in terms of luminance, contrast and structure similarity in the spatial domain to estimate the image quality. These structure-based metrics are also extended to multi-scale (MSSSIM) [7], complex domain (CW-SSIM) [8] and informationweighted (IW-SSIM) [9] versions. Instead of directly using the pixel va…”
Method
Metadata
- Authors
- Wang, Z., Simoncelli, E. P., & Bovik, A. C.
- Affiliations
- Courant Institute of Mathematical Sciences (US)
- The University of Texas at Austin (US)
- Venue
- The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003
- Publisher
- IEEE
- Published
- 2004-07-08
- Pages
- 1398-1402
- ISSN
- 1058-6393 (ISSN-L)
- OpenAlex
- W1580389772
- Cited by
- 6,253
- References
- 15
- Citations by year
- 2026: 344 · 2025: 675 · 2024: 729 · 2023: 668 · 2022: 662 · 2021: 641 · 2020: 563 · 2019: 438 · 2018: 305 · 2017: 267 · 2016: 206 · 2015: 173 · 2014: 166 · 2013: 112 · 2012: 89
- FWCI
- 3.98
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Asilomar Conference on Signals, Systems and Computers (ACSSC) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image and Video Quality Assessment
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Retrieval and Classification Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Keywords
- structural similarity, image quality assessment, multiscale analysis, human visual system, perceptual image quality, image generation, parameter calibration, viewing conditions
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.