The DOI matches its Crossref record; 81 of the 96 references it cites were found in a registry or an index; 6 of the 38 sources its claims cite were checked for support.
Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval
Open accessAbstract
This paper proposes a deep hashing framework, namely Unsupervised Deep Video Hashing (UDVH), for largescale video similarity search with the aim to learn compact yet effective binary codes. Our UDVH produces the hash codes in a self-taught manner by jointly integrating discriminative video representation with optimal code learning, where an efficient alternating approach is adopted to optimize the objective function. The key differences from most existing video hashing methods lie in 1) UDVH is an unsupervised hashing method that generates hash codes by cooperatively utilizing feature clustering and a specifically-designed binarization with the original neighborhood structure preserved in the binary space; 2) a specific rotation is developed and applied onto video features such that the variance of each dimension can be balanced, thus facilitating the subsequent quantization step. Extensive experiments performed on three popular video datasets show that UDVH is overwhelmingly better than the state-of-the-arts in terms of various evaluation metrics, which makes it practical in real-world applications.
Overview
CoChat Summary
TL;DR
The paper proposes Unsupervised Deep Video Hashing (UDVH), a self-taught deep hashing framework that jointly learns video representations and balanced binary codes via feature clustering, CCA, and a balanced rotation. On FCVID, YFCC, and ActivityNet, UDVH-TSN outperforms state-of-the-art video hashing baselines in retrieval accuracy and training efficiency.
Key findings
- UDVH-TSN achieves mAP@20 of 50.4% on FCVID at 128 bits, 4.8% higher than UDVH-LSTM (45.6%) and 8.9% over SSTH; on YFCC it reaches 29.8% vs. 26.1% and 24%; on ActivityNet it reaches 26.2% vs. 20.7% and 17.8%.
- Precision@100 at 64 bits is 51% on FCVID (at least about 3.6% higher than comparable baselines), 30.2% on YFCC vs. SSTH’s 26.3%, and 26.1% on ActivityNet vs. SSTH’s 20.4%.
- The balanced rotation (CCA-BR) outperforms CCA-ITQ and PCA-BR, producing flatter per-dimension variance; cluster number around 400 gives best mAP@20 at 128 bits (FCVID 0.504, YFCC 0.293, ActivityNet 0.262).
- Training is efficient: UDVH-TSN takes about 0.74–0.81 hours per loop and less than 4.5 hours total, versus over 8 hours for SSTH; balanced rotation costs only seconds.
Methods
- UDVH uses k-means on 10,000 sampled FC7 1024-d video features to generate C pseudo-labels, CCA to project features to k dimensions, and a balanced rotation solved as an orthogonality-constrained l2,1-norm maximization; it alternates updating R, B = sign(H × R), and deep parameters Θ via SGD minimizing ||F(Z; Θ) − B||_F^2, with t = 3–5 loops.
- The deep architecture uses modified Temporal Segment Networks (TSNs) with spatial and temporal ConvNets, fused global-pool features, and two fully connected layers (FC7 and FC8); 20 frames are uniformly sampled per video.
- Experiments use FCVID (91,223 videos, 239 categories; 45,611 train), YFCC (0.7M processed videos; 0.6M unlabeled train, 0.1M labeled test), and ActivityNet (~20,000 clips, 200 categories; ~10,000 train/test; 1,000 queries), comparing against AGH, SubMod, ITQ, SP, MFH, DH, SSTH, and UDVH-LSTM with mAP@K, PR, and Precision@100 at 16/32/64/128 bits.
Limitations
- The paper states ActivityNet gives the worst performance due to many untrimmed videos and large intra-class variance, with improvement there described as “a bit limited”; YFCC performance drops because many videos are low-quality amateur mobile clips; loss-function effects are not obvious (less than 3% improvement).
- Evident from the text: cluster number C, iteration count t, and other hyperparameters are chosen empirically, and the alternating optimization may only find a local optimum.
- Evident from the text: evaluation is limited to three video datasets and a selected set of baselines, and the method is unsupervised, relying on pseudo-labels rather than manual labels.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 8 Not valid · 22 Not checked · 75 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“In contrast to Principle Component Analysis (PCA), CCA is more effective in extracting discriminative information with the robustness in anti-noise [3], thus obtaining more discriminative video features after the projection.”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > A. Feature Embedding with Pseudo Labels
What the sources say
- Could not verifycites [3] sourceExperimental — accuracy not yet measured
Section III describes a kernelized correlation-filter tracker, and the Experimental Settings section says it uses HOG features. Searches for CCA, PCA, and their full names found no passage comparing their discriminative information or noise robustness. The copy therefore does not address the attributed claim. (LLM, conf=0.77)
plainstates a cause- 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.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Unfortunately, it has been acknowledged that the performance of those hashing schemes may degrade significantly if the projected dimensions are imbalanced [4], [6], [13], i.e., their variances vary a lot.”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > I. INTRODUCTION
What the sources say
- Not checked — no full text of the sourcecites [4] source
- Could not verifycites [6] sourceExperimental — accuracy not yet measured
Section 7.1 discusses compressing image features into short binary codes for retrieval, but does not address imbalanced projected dimensions or their effect on hashing performance. Searches for hashing, projected dimensions, and variance found no passage in our copy that makes the attributed claim. (LLM, conf=0.76)
- Could not verifycites [13] sourceExperimental — accuracy not yet measured
The Introduction (p. 1) discusses features that may be incompatible with hash coding, while Section 3.3 (p. 4) discusses redundancy among hash bits. Neither addresses whether unequal variances across projected dimensions significantly degrade hashing performance (searched: projected dimensions, imbalanced, variance). (LLM, conf=0.74)
hedgedstates a possible cause- 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.
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“More importantly, our optimization approach transforms the hash function learning into a regression problem with binary code B as the regression target, thus avoiding the use of a relaxation strategy to solve the nonconvex equation with binary constraints in most published works [3], [8], [18], [35], [48], [57], [58].”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > C. Objective Function and Optimization
What the sources say
- Could not verifycites [3] sourceExperimental — accuracy not yet measured
The cited paper’s Introduction and Weighted Interpolation Updating section describe a visual-tracking method, not hash-function learning. Searches for binary-code regression and relaxation of binary constraints found no passage addressing the point for which this paper is cited. (LLM, conf=0.76)
- Not checked — no full text of the sourcecites [8] source
- Not checked — no full text of the sourcecites [18] source
plainstates a cause- 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.
- Could not verify
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 results show that the combination of CCA and BR outperforms the other methods on the datasets substantially.”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > E. Results and Analysis 1) Parameter Analysis:
plainstates a cause- 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 states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Again, it is pretty clear that our BR scheme can always guarantee the flat variance of each dimension, regardless of which video features are used.”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > E. Results and Analysis 1) Parameter Analysis:
boostedstates a cause- 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.
53 more claims
Create a free account to see all 105 claims and every source verdict.
- It may state two separate claims as one. 42
- Probably not a substantial claim. 7
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > 5) Comparison with State-of-The-Arts:
“the mAP@20 values of UDVH-TSN (29.8%) and UDVH-LSTM (26.1%) at 128 bits”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > C. Implementation Details
“YFCC [64] 0.201 0.274 0.293 0.289 0.291”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > D. Complexity Analysis
“the time complexity is O(k 3 ) in (15) in solving B”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > C. Objective Function and Optimization
“B = sign(H × R), s.t. H ∈ R N ×k , R ∈ R k×k . (15”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > III. UNSUPERVISED DEEP VIDEO HASHING
“Those network parameters will be updated automatically during each iteration of hash function learning”
§ IEEE TRANSACTIONS ON IMAGE PROCESSING 1 Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval > C. Implementation Details
“the mini-batch back-propagation is only performed in the FC7 and FC8 layers”
Citations
Cites
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818
Representation Learning: A Review and New Perspectives
Bengio, Y., Courville, A., & Vincent, P. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.50 · Cited by 13,001
Two-Stream Convolutional Networks for Action Recognition in Videos
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1406.2199 · Cited by 6,756
The relationship between Precision-Recall and ROC curves
Davis, J., & Goadrich, M. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143874 · Cited by 6,559
Large-Scale Machine Learning with Stochastic Gradient Descent
Bottou, L. (2010) · Proceedings of COMPSTAT'2010 · doi:10.1007/978-3-7908-2604-3_16 · Cited by 5,795
Ucf101: A dataset of 101 human actions classes from videos in the wild
Soomro, K., Zamir, A. R., & Shah, M. (2012) · arXiv · doi:10.48550/arxiv.1212.0402 · Cited by 4,575
Approximate nearest neighbors: towards removing the curse of dimensionality
Indyk, P., & Motwani, R. (1998) · Proceedings of the thirtieth annual ACM symposium on Theory of computing - STOC '98 · doi:10.1145/276698.276876 · Cited by 4,188
Research works printed but not linked: 3
- over, 2014, enot, “TRECVID 2014—An overview of the goals, tasks, data, evaluation mechanisms and metrics, Proc TRECVID, 52
- xu, 2013, Harmonious hashing, Proc IJCAI, 1820
- liu, 2011, Hashing with graphs, Proc ICML, 1
Cited by
RGB-T Salient Object Detection via Fusing Multi-Level CNN Features
Zhang, Q., Huang, N., Yao, L., Zhang, D., Shan, C., & Han, J. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2959253 · Cited by 247
“classification [72], object tracking [73], [74] and instance-level object retrieval [75], [76].”
BackgroundSelf-citationLearning Modality-Consistency Feature Templates: A Robust RGB-Infrared Tracking System
Lan, X., Ye, M., Shao, R., Zhong, B., Yuen, P. C., & Zhou, H. (2019) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2019.2898618 · Cited by 124
“Efforts have been made in this field, and the decade has witnessed numerous tracking algorithms proposed to handle a variety of research issues [1]–[29].”
BackgroundRethinking Triplet Loss for Domain Adaptation
Deng, W., Zheng, L., Sun, Y., & Jiao, J. (2020) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2020.2968484 · Cited by 96
Adaptive Label Correlation Based Asymmetric Discrete Hashing for Cross-modal Retrieval
Li, H., Zhang, C., Jia, X., Gao, Y., & Chen, C. (2021) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2021.3102119 · Cited by 99
“More recently, the great success of deep learning has inspired the study of deep hashing methods [21], [22], [23], [24], [25], [26], [27].”
BackgroundDeep Salient Object Detection With Contextual Information Guidance
Liu, Y., Han, J., Zhang, Q., & Shan, C. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2930906 · Cited by 96
“Besides, we will apply our salient object detector to facilitate the representation ability of existing deep networks [73], [74] and real-world applications, including image retrieval [75], [76] and image classification [77].”
MethodSelf-citation
Metadata
- Authors
- Wu, G., Han, J., Guo, Y., Liu, L., Ding, G., Ni, Q., & Shao, L.
- Affiliations
- Lancaster University (GB)
- Tsinghua University (CN)
- Inception Institute of Artificial Intelligence (AE)
- Venue
- IEEE Transactions on Image Processing
- Published
- 2019-04
- Volume
- 28
- Issue
- 4
- Pages
- 1993-2007
- ISSN
- 1057-7149 (ISSN-L)
- 1941-0042
- OpenAlex
- W2900790473
- Cited by
- 148
- References
- 96
- Citations by year
- 2026: 2 · 2025: 9 · 2024: 9 · 2023: 21 · 2022: 24 · 2021: 22 · 2020: 29 · 2019: 32
- FWCI
- 10.38
- Type
- journal article
- Language
- en
- Open access
- green
- eprints.lancs.ac.uk
- Licence
- Locations
- IEEE Transactions on Image Processing (publishedVersion)
- PubMed (publishedVersion)
- Lancaster EPrints (Lancaster University) (acceptedVersion)
- Warwick Research Archive Portal (University of Warwick) (submittedVersion)
- Funders
- Royal Society (IE150997)
- Shenzhen Government (GJHZ20180419190732022)
- National Natural Science Foundation of China (61773301, 61571269)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Retrieval and Classification Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Video Analysis and Summarization
- Keywords
- video hashing, binary codes, feature clustering, binarization, quantization, alternating optimization
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
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).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval. https://app.cochat.ai/papers/10.1109/tip.2018.2882155
- The paper
- Wu, G., Han, J., Guo, Y., Liu, L., Ding, G., Ni, Q., & Shao, L. (2019). Unsupervised Deep Video Hashing via Balanced Code for Large-Scale Video Retrieval. IEEE Transactions on Image Processing. https://doi.org/10.1109/tip.2018.2882155
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.