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 1,001–1,100 of 6,047
Qin, Z., Xie, J., Lin, F., Huang, Y., & Shieh, H. D. (2017) · IEEE photonics journal · doi:10.1109/jphot.2017.2722000 · Cited by 37
“Therefore, we proposed a simulation method for see-through images based on diffraction theory, and revealed that MS-SSIM is the most suitable FR-IQA algorithm to assess the subjective quality of diffracted see-through images.”
MethodInfluentialMRI and PET image fusion using structure tensor and dual ripplet-II transform
Shahdoosti, H. R., & Mehrabi, A. (2017) · Multimedia Tools and Applications · doi:10.1007/s11042-017-5067-1 · Cited by 40
“ASSIM calculates the SSIM between Fk and Pk and that between Fk and Mh, and then averages these values; the SSIM index considers the contrast, structure, and luminance comparison measures [38].”
MethodImage Visual Realism: From Human Perception to Machine Computation
Fan, S., Ng, T.-T., Koenig, B. L., Herberg, J. S., Jiang, M. X., Shen, Z., & Zhao, Q. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2747150 · Cited by 45
“For automated prediction, reference-based image quality metrics (IQMs) such as mean squared error [9] and structural similarity index [10] are used to quantitatively calculate the distortions induced by global illumination and artifacts given an ideal reference image [1], [2], [3], [11].”
MethodReal-Time Data Analysis and Autonomous Steering of Synchrotron Light Source Experiments
Biçer, T., Gürsoy, D., Kettimuthu, R., Foster, I. T., Ren, B., Andrede, V. D., & Carlo, F. D. (2017) · IEEE International Conference on eScience · doi:10.1109/escience.2017.53 · Cited by 46
“We use the Multi-Scale Structural Similarity Index (MSSSIM) [8] to compute the similarity score.”
MethodAn objective assessment method based on multi-level factors for panoramic videos
Yang, S., Zhao, J., Jiang, T., Wang, J., Rahim, T., Zhang, B., Xu, Z., & Fei, Z. (2017) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip.2017.8305133 · Cited by 37
Blind Stereoscopic Image Quality Assessment Based on Hierarchical Learning
Liu, T.-J., Lin, C.-T., Liu, H., et al. (2019) · IEEE Access · doi:10.1109/access.2018.2890304 · Cited by 37
“The first type is the application of two-dimensional (2D) image quality assessment to evaluate the quality scores of stereoscopic images, such as VIF [3], [4], PSNR, SSIM [5], MS-SSIM [6] and so on.”
BackgroundSmart electronic gastroscope system using a cloud–edge collaborative framework
Ding, S., Li, L., Li, Z., Wang, H., & Zhang, Y. (2019) · Future Generation Computer Systems · doi:10.1016/j.future.2019.04.031 · Cited by 39
An End-to-End Deep Learning Image Compression Framework Based on Semantic Analysis
Wang, C., Han, Y., & Wang, W. (2019) · Applied Sciences · doi:10.3390/app9173580 · Cited by 39
“We performed 0–30 iterative compressions on the 24 images in the Kodak dataset and calculated the corresponding MS-SSIM [22] values to verify the impact of the number of iterations on the quality of the recovered images.”
MethodMarstaller, J., Tausch, F., & Stock, S. (2019) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2019.00036 · Cited by 45
“Additionally, we tracked Multiscale Structural Similarity (MS-SSIM) for the image reconstructions of the decoder during training based on Wang et al. [34].”
MethodCai, J., Cao, Z., & Zhang, L. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2963956 · Cited by 45
Blind Quality Assessment for Tone-Mapped Images by Analysis of Gradient and Chromatic Statistics
Fang, Y., Yan, J., Du, R., Zuo, Y., Wen, W., Zeng, Y., & Li, L. (2020) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2020.2991528 · Cited by 37
“According to the amount of information used in the tasks, image quality assessment (IQA) methods can be classified into full-reference [9], [10], reduced-reference [11], [12] and no-reference approaches [13]–[17].”
BackgroundCai, J., Meng, Z., & Ho, C. M. (2020) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw50498.2020.00235 · Cited by 39
“Besides, the multi-scale structural similarity (MS-SSIM) [34] supplies more flexibility than traditional SSIM in incorporating the variations of viewing conditions.”
Performances of Polarization-Retrieve Imaging in Stratified Dispersion Media
Wang, X., Hu, T., Li, D., et al. (2020) · Remote Sensing · doi:10.3390/rs12182895 · Cited by 37
Improved Noise2Noise Denoising with Limited Data
Calvarons, A. F. (2021) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw53098.2021.00089 · Cited by 46
“This loss is the combination of the Huber loss (as the Moreau envelope of non-differentiable l(1) norm [25]) and MS-SSIM [41] metric.”
MethodRegion-Adaptive Deformable Network for Image Quality Assessment
Shi, S., Bai, Q., Cao, M., Xia, W., Wang, J., Chen, Y., & Yang, Y.-J. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw53098.2021.00042 · Cited by 30
“FR-IQA methods commonly include SSIM [30], MS-SSIM [31] and PSNR, etc .”
MethodInfluentialEnd-to-End Neural Video Coding Using a Compound Spatiotemporal Representation
Liu, H., Lu, M., Chen, Z., Cao, X., Ma, Z., & Wang, Y. (2022) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2022.3150014 · Cited by 47
Lin, C., Wu, J., Chen, P.-Y., Lai, H., Li, C., Kuo, C., & Pai, N. (2021) · IEEE Access · doi:10.1109/access.2021.3107608 · Cited by 40
“Thus, d 1 , d 2 , and d 3 are also small constants; the values of parameters α , β , and γ are set to 1 [42].”
BackgroundMethodEPES: Point Cloud Quality Modeling Using Elastic Potential Energy Similarity
Xu, Y., Yang, Q., Yang, L., et al. (2021) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2021.3114510 · Cited by 36
Neural Video Compression Using GANs for Detail Synthesis and Propagation
Mentzer, F., Agustsson, E., Ballé, J., Minnen, D. C., Johnston, N., & Toderici, G. D. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19809-0_32 · Cited by 76
“In order to assess how quantitative metrics predict the user study, we employ the well-known PSNR and MSSSIM [42].”
Two-Stage Octave Residual Network for End-to-End Image Compression
Chen, F., Xu, Y., & Wang, L. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i4.20308 · Cited by 39
A Novel Image Steganographic Approach for Hiding Text in Color Images using HSI Color ModelPreprint
Muhammad, K., Ahmad, J., Farman, H., & Zubair, M. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1503.00388 · Cited by 49
End-to-end Piece-wise Unwarping of Document Images
Das, S., Singh, K. Y., Wu, J., Bas, E., Mahadevan, V., Bhotika, R., & Samaras, D. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00423 · Cited by 58
“Image similarity metric, MS-SSIM [32] is based on local image statistics (mean and variance) of the unwarped and scanned (ground-truth) images calculated over multiple Gaussian pyramid scales.”
MethodInfluentialFast DNA-PAINT imaging using a deep neural network
Narayanasamy, K. K., Rahm, J. V., Tourani, S., & Heilemann, M. (2022) · Nature Communications · doi:10.1038/s41467-022-32626-0 · Cited by 41
Learning Spatiotemporal Interactions for User-Generated Video Quality Assessment
Zhu, H., Chen, B., Zhu, L., & Wang, S. (2022) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2022.3207148 · Cited by 42
“The multi-scale processing scheme also matches the HVS mechanism [24], [61], [62], providing the more stable and accurate quality modeling.”
MethodResultCLSA: A Contrastive Learning Framework With Selective Aggregation for Video Rescaling
Tian, Y., Yan, Y., Zhai, G., Chen, L., & Gao, Z. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2023.3242774 · Cited by 42
“Following previous works [86], [93] on learnable video compression, we adopt PSNR and MS-SSIM [117] to evaluate the distortion between the reconstructed video and the original frame.”
MethodToward Fashion Intelligence in the Big Data Era: State-of-the-Art and Future Prospects
Liu, L., Zhang, H., Zhou, D., & Shi, J. (2023) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2023.3285880 · Cited by 41
“Quantitative evaluation metrics on image quality and diversity are generally inception score [205] and Multi-Scale Structural Similarity (MS-SSIM) [206].”
BackgroundUngureanu, V.-I., Negȋrla, P., & Korodi, A. (2024) · Sensors · doi:10.3390/s24030791 · Cited by 36
“The formula used to find SSIM for two windows x and y of common size N × N is defined in (5) [15]: where µ x is the pixel-sample mean of window x; µ y is the pixel-sample mean of window y; σ 2x is the variance of window x; σ 2y is the variance of window y; σ xy is the covariance of the two windows;…”
MethodQu, Q., Liang, H., Chen, X., Chung, Y. Y., & Shen, Y. (2024) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2024.3372037 · Cited by 52
“Our benchmarking considered prevalent full-reference image quality assessment metrics (FR-IQA) such as PSNR, SSIM [56], MS-SSIM [58], IW-SSIM [57], VIF [44], FSIM [63], GMSD [61], VSI [62], DSS [2], HaarPSI [41], MDSI [33], LPIPS [64], PieAPP [37], and DISTS [9], along with no-reference image…”
MethodPerceptual Information Fidelity for Quality Estimation of Industrial Images
Gu, K., Liu, H., Liu, Y., Qiao, J., Zhai, G., & Zhang, W. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3454160 · Cited by 42
“With the increasingly deep studies on the HVS, in [10] and [11], Wang et al. developed the structural similarity (SSIM) and its modified version multi-scale SSIM (MS-SSIM) to estimate the image quality perceived by humans.”
GauFRe: Gaussian Deformation Fields for Real-Time Dynamic Novel View Synthesis
Liang, Y., Khan, N., Li, Z., Nguyen-Phuoc, T., Lanman, D. R., Tompkin, J., & Xiao, L. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00262 · Cited by 102
Xu, C., Mashhadi, M. B., Ma, Y., Tafazolli, R., & Wang, J. (2025) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2025.3559120 · Cited by 61
“…as non-distribution-based perceptual distances using the contrastive language-image pre-training (CLIP) similarity [30] or multi-scale structural similarity (MS-SSIM) [31], which are expressed as: where CLIP( X , ˆ X ) and MS - SSIM( X , ˆ X ) represent the CLIP and MS-SSIM similarity functions.”
Perceptual Depth Quality in Distorted Stereoscopic Images
Wang, J., Wang, S., Ma, K., & Wang, Z. (2016) · IEEE Transactions on Image Processing · doi:10.1109/tip.2016.2642791 · Cited by 34
Self-citationFull-reference image quality assessment based on image segmentation with edge feature
Shi, Z., Zhang, J., Cao, Q., Pang, K., & Luo, T. (2017) · Signal Processing · doi:10.1016/j.sigpro.2017.11.015 · Cited by 36
Borra, S., Thanki, R. M., Dey, N., & Borisagar, K. R. (2018) · Smart Health · doi:10.1016/j.smhl.2018.02.001 · Cited by 36
Perceptual image quality assessment through spectral analysis of error representations
Temel, D., & AlRegib, G. (2018) · Signal Processing Image Communication · doi:10.1016/j.image.2018.09.005 · Cited by 39
Embedded deep learning in ophthalmology: making ophthalmic imaging smarter
Teikari, P., Najjar, R. P., Schmetterer, L. F., & Miléa, D. (2019) · Therapeutic Advances in Ophthalmology · doi:10.1177/2515841419827172 · Cited by 43
Open Questions about Generative Adversarial Networks
Odena, A. (2019) · Distill · doi:10.23915/distill.00018 · Cited by 45
Rahnemoonfar, M., Dobbs, D., Yari, M., & Starek, M. J. (2019) · Remote Sensing · doi:10.3390/rs11091128 · Cited by 39
“This results in a small error value in every pixel which is even more pronounced when comparing SSIM over other metrics.”
MethodInfluentialGEANT4 Monte Carlo simulations for virtual clinical trials in breast X-ray imaging: Proof of concept
di Franco, F., Sarno, A., Mettivier, G., Hernandez, A. M., Bliznakova, K. S., Boone, J. M., & Russo, P. (2020) · Physica Medica · doi:10.1016/j.ejmp.2020.05.007 · Cited by 37
Geometric rectification of document images using adversarial gated unwarping network
Liu, X., Meng, G., Fan, B., Xiang, S., & Pan, C. (2020) · Pattern Recognition · doi:10.1016/j.patcog.2020.107576 · Cited by 34
Sun, S., Mu, L., Wang, L., Liu, P., Liu, X., & Zhang, Y. (2021) · Remote Sensing · doi:10.3390/rs13030475 · Cited by 40
“The structural features of the image signal can present a similarity in more specific situations [43].”
BackgroundEvent-based Synthetic Aperture Imaging with a Hybrid Network
Zhang, X., Wei, L., Yu, L., Yang, W. T., & Xia, G.-S. (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/cvpr46437.2021.01401 · Cited by 38
“In the dense occlusion experiment, the metrics PSNR and SSIM [22] are employed for quantitative comparison, where the aligned APS images captured by DAVIS346 are considered as the ground truth.”
MethodInfluentialDeep Perceptual Preprocessing for Video Coding
Chadha, A., & Andreopoulos, Y. (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/cvpr46437.2021.01461 · Cited by 43
“3, where DVC is very substantially outperformed in terms of bitrate vs. PSNR and MS-SSIM (the metrics used in their work) by both DPP+AVC/H.264 and DPP+HEVC/H.265 under our encoding recipe.”
BackgroundInfluential‘CADSketchNet’ - An Annotated Sketch dataset for 3D CAD Model Retrieval with Deep Neural Networks
Manda, B., Dhayarkar, S., Mitheran, S., Viekash, V. K., & Muthuganapathy, R. (2021) · Computers & Graphics · doi:10.1016/j.cag.2021.07.001 · Cited by 38
Zhu, M., Hou, G., Chen, X., Xie, J., Lu, H., & Che, J. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00222 · Cited by 48
Real-Time Depth Video-Based Rendering for 6-DoF HMD Navigation and Light Field Displays
Bonatto, D., Fachada, S., Rogge, S., Munteanu, A., & Lafruit, G. (2021) · IEEE Access · doi:10.1109/access.2021.3123529 · Cited by 37
“In the first experiment we compute the Peak Signal to Noise Ratio (PSNR) and Multi-Scale Structural Similarity Index Measure (MS-SSIM) [78] of our view synthesis against images of Toystable (Plane A), Museum, Fencing and Painter natural datasets for one to eight input views (see Figure 16).”
MethodResultInfluentialLearning Super-Resolution Reconstruction for High Temporal Resolution Spike Stream
Xiang, X., Zhu, L., Li, J., Wang, Y., Huang, T., & Tian, Y. (2021) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2021.3130147 · Cited by 45
“We used the Learned Perceptual Image Patch Similarity (LPIPS) [47] for high level features metrics as a fraction between 0 (high similarity) and 1 (low similarity) and the Structural Similarity (SSIM) [48] as a fraction between 0 (low similarity) and 1 (high similarity).”
MethodLiao, X.-R., Wei, X.-K., Zhou, M.-L., & Kwong, S. (2024) · IEEE transactions on broadcasting · doi:10.1109/tbc.2023.3294835 · Cited by 33
Megahed, M., & Mohammed, A. (2023) · Wiley Interdisciplinary Reviews Computational Statistics · doi:10.1002/wics.1629 · Cited by 44
“…have been driven from SSIM to add different evaluation criteria like multi-scale structural similarity index (MS-SSIM; Rouse & Hemami, 2008; Z. Wang et al., 2003), information content weighted structural similarity index (IW-SSIM; Z. Wang & Li, 2011), and part-based SSIM (PSSIM; Shi et al.,…”
MethodResultHARP: Personalized Hand Reconstruction from a Monocular RGB Video
Karunratanakul, K., Prokudin, S., Hilliges, O., & Tang, S. (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/cvpr52729.2023.01231 · Cited by 61
“LPIPS ↓ MS-SSIM ↑ METRO [36] 0.651 - - - S2Hand [8] 0.430 0.058 0.270 0.595 HTML [59] 0.778 0.033 0.136 0.791 NHA [19] 0.860 0.025 0.114 0.878 NIMBLE [35] 0 suggest that HARP can faithfully reconstruct hand appearance with much higher details than the baselines.”
MethodInfluentialSpatial landmark detection and tissue registration with deep learning
Ekvall, M., Bergenstråhle, L., Andersson, A., Czarnewski, P., Olegård, J., Käll, L., & Lundeberg, J. (2024) · Nature Methods · doi:10.1038/s41592-024-02199-5 · Cited by 35
Preprocessing Enhanced Image Compression for Machine Vision
Lu, G., Ge, X., Zhong, T., Hu, Q., & Geng, J. (2022) · IEEE transactions on circuits and systems for video technology (Print) · doi:10.1109/tcsvt.2024.3441049 · Cited by 67
“Most existing image compression methods [3, 8, 9, 39] aim to reduce reconstruction distortion in terms of the human visual system and are optimized based on the pixel field metrics such as PSNR or MS-SSIM [41].”
MethodModels of Monocular and Binocular Visual Perception in Quality Assessment of Stereoscopic Images
Shao, F., Lin, W., Jiang, G., & Dai, Q.-H. (2016) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2016.2538720 · Cited by 32
SSIM-Motivated Two-Pass VBR Coding for HEVC
Wang, S., Rehman, A., Zeng, K., Wang, J., & Wang, Z. (2016) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2016.2580398 · Cited by 36
“For fixed MS-SSIM, the rate reductions for LDB and RA cases are 14.7% and 7.6%, respectively.”
MethodInfluentialSelf-citationImage Quality Assessment Using Directional Anisotropy Structure Measurement
Ding, L., Huang, H., & Zang, Y. (2017) · IEEE Transactions on Image Processing · doi:10.1109/tip.2017.2665972 · Cited by 36
Multisensor X-ray inspection of internal defects in horticultural products
van Dael, M., Verboven, P., Dhaene, J., Hoorebeke, L. V., Sijbers, J., & Nicolaï, B. M. (2017) · Postharvest Biology and Technology · doi:10.1016/j.postharvbio.2017.02.002 · Cited by 39
Collaborative Representation Cascade for Single-Image Super-Resolution
Zhang, Y., Zhang, Y., Zhang, J., Xu, D., Fu, Y., Wang, Y., Ji, X., & Dai, Q. (2017) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2017.2705480 · Cited by 34
“SSIM) [54], visual information fidelity (VIF) [55], and feature similarity (FSIM) [56].”
BackgroundSpatial-spectral cube matching frame for spectral CT reconstruction
Wu, W., Zhang, Y., Wang, Q., Liu, F., Luo, F., & Yu, H. (2018) · Inverse Problems · doi:10.1088/1361-6420/aad67b · Cited by 39
“These conclusions are also confirmed with the indexes of PSNR, FSIM and SSIM. b) Material Decomposition—”
MethodInfluentialStructural Similarity Index with Predictability of Image Blocks
Ponomarenko, N. N., Egiazarian, K., Лукин, В. В., & Abramova, V. (2018) · International Conference on Mathematical Methods in Electromagnetic Theory · doi:10.1109/mmet.2018.8460285 · Cited by 36
“However, this indirect way of accounting of masking effects (it is impossible to regulate the value of these effects) alongside with neglecting peculiarities of color perception by HVS has led to the fact that, according to data for the database TID2013 [4], the metrics SSIM and MS-SSIM perform worse than many objecti…”
MethodInfluentialNaturalization Module in Neural Networks for Screen Content Image Quality Assessment
Chen, J., Shen, L., Zheng, L., & Jiang, X. (2018) · IEEE Signal Processing Letters · doi:10.1109/lsp.2018.2871250 · Cited by 33
“It is evident that traditional approaches such as MS-SSIM [13] and VIF [14] have high accuracies for natural images but perform less optimally on SCIs.”
MethodNeural Multi-scale Image Compression
Nakanishi, K. M., Maeda, S., Miyato, T., & Okanohara, D. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-20876-9_45 · Cited by 41
Super Resolution of B-Mode Ultrasound Images With Deep Learning
Temiz, H., & Bılge, H. Ş. (2020) · IEEE Access · doi:10.1109/access.2020.2990344 · Cited by 38
“…quality metrics: peak signal to noise ratio (PSNR), structural similarity index (SSIM) [58], multi-scale structural similarity index (MS-SSIM) [59], feature similarity index (FSIM) [60], mean absolute deviation (MAD), Haar wavelet-based perceptual similarity index (HaarPSi) [61], gradient…”
BackgroundImproving Psnr-Based Quality Metrics Performance For Point Cloud Geometry
Javaheri, A., Brites, C., Pereira, F., et al. (2020) · Proceedings - International Conference on Image Processing · doi:10.1109/icip40778.2020.9191233 · Cited by 35
“cultural heritage), however further optimizations are possible such as using multiple distances (or scales) as other 2D quality metrics such as MS-SSIM [18].”
BackgroundAdversarial Privacy-preserving Filter
Zhang, J., Sang, J., Zhao, X., Huang, X., Sun, Y., & Hu, Y. (2020) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3394171.3413906 · Cited by 57
“03 by default [33].”
Quality Assessment for Omnidirectional Video: A Spatio-Temporal Distortion Modeling Approach
Gao, P., Zhang, P., & Smolić, A. (2020) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2020.3044458 · Cited by 43
“…for quality assessment of still images which estimates perceptual distortions by considering structural information, and its extension MultiScale-SSIM (MS-SSIM) [4] provides more flexibility than the single-scale version by incorporating the variations of image resolution and viewing conditions.”
BackgroundBuilding Complex Seismic Velocity Models for Deep Learning Inversion
Ren, Y., Nie, L., Yang, S., Jiang, P., & Chen, Y. (2021) · IEEE Access · doi:10.1109/access.2021.3051159 · Cited by 39
“Similarly to the SeisInvNet paper, the loss function is defined based on the L 2 norm and MSSIM [45] of the predicted and the true velocity.”
MethodSemantic Perceptual Image Compression With a Laplacian Pyramid of Convolutional Networks
Wang, J., Duan, Y., Tao, X., Xu, M., & Lü, J. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3065244 · Cited by 40
“Structural similarity index (SSIM) [56] and multi-scale structure similarity index 1 http://r0k.us/graphics/kodak/ (MS-SSIM) [57] are also well-established metrics for evaluating the lossy image compression algorithms.”
MethodImage scrambling adversarial autoencoder based on the asymmetric encryption
Bao, Z., Xue, R., & Jin, Y. (2021) · Multimedia Tools and Applications · doi:10.1007/s11042-021-11043-3 · Cited by 41
“[9] Ours SSIM (encrypted) 0.0379 0.0073 0.6308 0.01 0.0107 MS-SSIM (encrypted) 0.0847 0.1376 0.3735 / 0.0582 ER (encrypted) 3.0738 0.2924 1.5165 / 4.4662 EDR (encrypted) 0.8606 0.9870 0.4305 / 0.8833 PSNR (decrypted) 100 14.9573 29.7823 37.43 27.5087 SSIM (decrypted) 1.0000 0.6324 0.9409 0.93 0.9115 MS-SSIM (decrypted…”
MethodInfluentialComparative study of the methodologies used for subjective medical image quality assessment
Lévêque, L., Outtas, M., Liu, H., & Zhang, L. (2021) · Physics in Medicine and Biology · doi:10.1088/1361-6560/ac1157 · Cited by 46
Learning Context-Based Nonlocal Entropy Modeling for Image Compression
Li, M., Zhang, K., Li, J., Zuo, W., Timofte, R., & Zhang, D. D. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2021.3104974 · Cited by 51
Multi-Source Fusion and Automatic Predictor Selection for Zero-Shot Video Object Segmentation
Zhao, X., Pang, Y., Yang, J., Zhang, L., & Lu, H. (2021) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3474085.3475192 · Cited by 46
“For depth stream, we follow the related works [23, 31, 34] to adopt the combination loss of L1 and SSIM [49].”
Joint Geometry and Color Projection-Based Point Cloud Quality Metric
Javaheri, A., Brites, C., Pereira, F. M. B., et al. (2022) · IEEE Access · doi:10.1109/access.2022.3198995 · Cited by 35
“In [37], VIFP, SSIM, MS-SSIM and Y-PSNR are used, while, in [11], only Y-PSNR is used.”
BackgroundInfluentialQuantitative image quality evaluation of MR images using perceptual difference models
Miao, J., Huo, D., & Wilson, D. L. (2008) · Medical Physics · doi:10.1118/1.2903207 · Cited by 35
Extracting a Good Quality Frontal Face Image From a Low-Resolution Video Sequence
Nasrollahi, K., & Moeslund, T. B. (2011) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2011.2162267 · Cited by 35
An information-theoretic analysis of deep latent-variable modelsPreprint
Alemi, A. A., Poole, B., Fischer, I., Dillon, J. V., Saurous, R. A., & Murphy, K. P. (2018) · arXiv (Cornell University) · Cited by 48
Crowd Counting with Deep Structured Scale Integration NetworkPreprint
Liu, L., Qiu, Z., Li, G., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.08692 · Cited by 35
“We ameliorate the original MS-SSIM [39] with dilation operations to enlarge the diversity of the sizes of local regions and force our network to capture the local correlation within regions of different sizes.”
BackgroundMethodInfluentialDiffusion Probabilistic Models beat GANs on Medical ImagesPreprint
Müller‐Franzes, G. A., Niehues, J., Khader, F., et al. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2212.07501 · Cited by 35
“Third, the Multiscale Structural Similarity Index Measure (MSSSIM) [27] that is a generalized version of the SSIM [28] by applying SSIM at different resolutions.”
MethodAhuja, V. R., Gupta, U., Rapole, S. R., Saxena, N., Hofmann, R., Day-Stirrat, R. J., Prakash, J., & Yalavarthy, P. K. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3172211 · Cited by 34
“…are good at obtaining a high Peak Signal to Noise Ratio (PSNR), they tend to compromise on the high-frequency details like texture since they use generic optimization functions based on Mean Squared Error (MSE) in the image space which are defined based on the pixel-wise differences [14]–[16].”
BackgroundEdge-Guided Hyperspectral Image Compression With Interactive Dual Attention
Guo, Y., Tao, Y., Chong, Y., et al. (2022) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2022.3233375 · Cited by 34
“The details are defined in [31].”
BackgroundMethodUIESC: An Underwater Image Enhancement Framework via Self-Attention and Contrastive Learning
Chen, R., Cai, Z., & Yuan, J. (2023) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2023.3249794 · Cited by 37
“In addition, we adopt MS-SSIM [37] loss to keep more high-frequency information of image.”
MethodRestoration and enhancement on low exposure raw images by joint demosaicing and denoising
Ma, J., Wang, G., Zhang, L., & Zhang, Q. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.03.018 · Cited by 39
VNVC: A Versatile Neural Video Coding Framework for Efficient Human-Machine Vision
Sheng, X., Li, L., Liu, D., & Li, H. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3356548 · Cited by 53
“When using 1 − MS-SSIM [74] as the distortion, each λ value is divided by 50.”
MethodATGAN: A SAR Target Image Generation Method for Automatic Target Recognition
Zeng, Z., Tan, X., Zhang, X., Huang, Y., Wan, J., & Chen, Z. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3370185 · Cited by 39
Diffusion in the Dark: A Diffusion Model for Low-Light Text Recognition
Nguyen, C. M., Chan, E. R., Bergman, A. W., & Wetzstein, G. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00410 · Cited by 48
“It is well known that MSE, and therefore PSNR, cannot capture perceptual similarities [24, 80, 79, 78, 69, 18].”
BackgroundESVO2: Direct Visual-Inertial Odometry With Stereo Event Cameras
Niu, J., Zhong, S., Lu, X., Shen, S., Gallego, G., & Zhou, Y. B. (2025) · IEEE Transactions on Robotics · doi:10.1109/tro.2025.3548523 · Cited by 68
“II, the proposed AA representation leads to the highest score using the multi-component structure similarity index measure (3-SSIM) [74] as evaluation metric.”
A novel methodology for mapping objective video quality metrics to the subjective MOS scale
Moldovan, A.-N., Ghergulescu, I., & Muntean, C. H. (2014) · IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB) · doi:10.1109/bmsb.2014.6873572 · Cited by 38
“On the overall data the MS-SSIM metric performs only slightly worse, while the PSNR metric had the lowest performance among the considered metrics.”
MethodInfluentialQuality-of-Experience of Adaptive Video Streaming
Duanmu, Z., Ma, K., & Wang, Z. (2017) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3123266.3123418 · Cited by 33
“First, from the significant improvement of MS-SSIM and SSIMplus upon SSIM, we may conclude that the multi-scale approach performs better against variations in resolution.”
MethodInfluentialSelf-citationImage-Dependent Local Entropy Models for Learned Image Compression
Minnen, D. C., Toderici, G. D., Singh, S., Hwang, S. J., & Covell, M. (2018) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2018.8451502 · Cited by 48
“EMPIRICAL EVALUATION The network architecture is the same for all experiments, but we trained two versions of the baseline model: one using mean squared error (MSE) and one using MS-SSIM as the objective function.”
MethodInfluentialAutomatic ISP Image Quality Tuning Using Nonlinear Optimization
Nishimura, J., Gerasimow, T., Rao, S., et al. (2018) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2018.8451818 · Cited by 34
“all zipper artifact and false color at high frequency (as indicated in red arrow on OV16860-ISO50) is properly addressed. Table 1 shows Mean Absolute Difference (MAD) in 8bit scale, SSIM, and MS-SSIM [10] between reference and different tunings of each key ISP block. ‘Auto-tuned’ provides significant IQ improvement ov…”
BackgroundObject-Centered Image Stitching
Herrmann, C., Wang, C., Bowen, R. S., et al. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01219-9_50 · Cited by 34
“We provide some examples and their evaluation values (maximum MS-SSIM with direct object comparison) in the captions of the figures above.”
MethodInfluentialImage Denoising Using a Deep Encoder-Decoder Network with Skip Connections
Couturier, R., Perrot, G., & Salomon, M. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-04224-0_48 · Cited by 39
“Such an investigation can be found in [22], a paper in which the authors compared several losses considering two state-of-the-art metrics for image quality: the Structural SIMilarity (SSIM) index [16] and the MultiScale Structural SIMilarity (MS-SSIM) index [17].”
BackgroundCrowd Counting via Multi-view Scale Aggregation Networks
Qiu, Z., Liu, L., Li, G., Wang, Q., Xiao, N., & Lin, L. (2019) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2019.00259 · Cited by 29
“Finally, the MS-SSIM loss is defined as following: (5) where the exponents α j , β j and γ j are used to adjust the relative importance of different components and they are set with the same values as in [4].”
BackgroundMethodInfluentialModeling the Screen Content Image Quality via Multiscale Edge Attention Similarity
Yang, Q., Ma, Z., Xu, Y., Yang, L., Zhang, W., & Sun, J. (2019) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2019.2954063 · Cited by 34
“Multiple IQA models are compared against the proposed MSEA similarity index, including PSNR, SSIM [21], MSSIM [20], IWSSIM [22], VIF [24], IFC [25], FSIM [23], SCQI [61], SIQM [5], SQI [2], ESIM [7], MDOGS [19], GFM [18], where the last five IQA models are specifically designed for SCIs, while others…”
BackgroundMethodUser-Generated Video Quality Assessment: A Subjective and Objective Study
Li, Y., Meng, S., Zhang, X., Wang, M., Wang, S., Wang, Y., & Ma, S. (2021) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2021.3122347 · Cited by 39
“clude reference-based models PSNR, SSIM, MS-SSIM [56], VIF [57], SpEED-QA, ViS3 and VMAF, as well as NR models BRISQUE [58], NIQE [59], VBLIINDS and VIIDEO.”
Pashaei, M., Starek, M. J., Kamangir, H., & Berryhill, J. (2020) · Remote Sensing · doi:10.3390/rs12111757 · Cited by 39
“Similar to the human visual system, which is highly adapted for extracting structural information from the viewing scene, SSIM index provides a perceptual metric that quantifies image quality degradation based on perceived image quality [68].”
BackgroundSemantic Example Guided Image-to-Image Translation
Huang, J., Liao, J., & Kwong, S. (2020) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2020.3001536 · Cited by 35
“According to [47], the LPIPS metric can outperform lots of perceptual metrics like PSNR [16], and SSIM [40], as it provides an emergent embedding which agrees surprisingly well with humans.”
BackgroundA Spatial RNN Codec for End-to-End Image Compression
Lin, C., Yao, J., Chen, F., et al. (2020) · 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/cvpr42600.2020.01328 · Cited by 30
“The RD curves of different methods for MS-SSIM. learning based models are compared with ou model.”
MethodInfluentialDeep Residual Autoencoders for Expectation Maximization-Inspired Dictionary Learning
Tolooshams, B., Dey, S., & Ba, D. E. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.3005348 · Cited by 39
CoSinGAN: Learning COVID-19 Infection Segmentation from a Single Radiological Image
Zhang, P., Zhong, Y., Deng, Y., Tang, X., & Li, X. (2020) · Diagnostics · doi:10.3390/diagnostics10110901 · Cited by 43
“Different from mean-based metrics like L1 distance and MSE, SSIM [36] and MS-SSIM [34] are perceptually motivated metrics to evaluate image similarity based on local structure.”
BackgroundMethodStarfish: resilient image compression for AIoT cameras
Hu, P., Im, J., Asgar, Z., & Katti, S. (2020) · ACM Conference on Embedded Networked Sensor Systems (SenSys) · doi:10.1145/3384419.3430769 · Cited by 53
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.