Pith. sign in

REVIEW 4 major objections 5 minor 44 references

Video Affective Effects Prediction with Multi-modal Fusion and Shot-Long Temporal Context

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims a multi-modal video emotion framework with per-task modality selection, two-time-scale temporal fusion, and residual-based progressive training beats prior methods by a large margin on LIRIS-ACCEDE.

desk verdict A reasonable engineering pipeline whose headline 'large margin' claim is undercut by test-set selection and by the paper's own arousal MSE, which is worse than a cited baseline. read the letter →

arxiv 1909.01763 v1 pith:IUXU5D7M submitted 2019-09-01 cs.CV cs.LG

Jie Zhang , Yin Zhao , Longjun Cai , Chaoping Tu , Wu Wei This is my paper · ORCID
classification cs.CVcs.LG
keywords affectivevideocontentanalysisvalenceandarousalpredictionmulti-modalfusiontemporalcontextmodelingLSTMprogressivetrainingLIRIS-ACCEDE
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is trying to establish that predicting a video's continuous emotional effect—its valence and arousal values—gets substantially better when three design decisions are made together. First, each emotion dimension gets its own modality set, chosen by measuring how informative audio, scene, facial-expression, and action features are for that dimension. Second, multi-modal features are fused with a residual-based progressive training rule: modalities are added one at a time, and each new modality is trained only to supply what the already-fused features are missing. Third, temporal context is handled at two time scales, with short in-clip LSTMs per modality and a longer clip-level context that is modeled differently for valence and arousal. On the LIRIS-ACCEDE benchmark the complete pipeline reports valence MSE of 0.071 and arousal PCC of 0.419, better than the state-of-the-art systems it compares with, which is the concrete payoff the paper is arguing for.

What carries the argument

The load-bearing mechanism is the residual-based progressive training strategy. Modalities are first trained individually, ranked by validation performance, and then added one by one: at step i the already-fused features $f_{i-1}$ are held fixed, and the new modality's LSTMs are trained to fit the residual mapping $H_i(x) = f_i - f_{i-1}$, so an uninformative modality is pushed to learn a near-zero mapping while an informative one fills in missing feature structure. Around this sits the two-time-scale temporal structure: bidirectional LSTMs inside each clip produce modality-specific clip embeddings, which are summed for the clip-level multi-modal feature, and then a second LSTM layer over clips models long context for valence while an exponential moving average post-processing models the slow arousal trend. This machinery is what the paper credits with avoiding the overfitting of joint feature-level training and the information loss of decision-level fusion.

What would settle it

Rerun the same framework on LIRIS-ACCEDE while fixing beta, clip count, and modality sets using only a validation split before any test labels are seen; if valence MSE rises above 0.071 or arousal PCC falls below 0.419, part of the reported margin over the state of the art is an artifact of test-set selection.

Watch

Extended reading notes

Core claim

The central claim is that the right decomposition of video affect prediction makes previously reported numbers look weak: per-task modality selection (audio, scene, and expression for valence; audio, scene, and action for arousal), two-stage temporal fusion, and residual-based progressive training combine into a framework whose valence and arousal predictions on LIRIS-ACCEDE beat the listed prior methods by a large margin. The paper also claims that audio is the most informative single modality for both dimensions, that expressions matter for valence but not arousal, that actions matter for arousal but not valence, and that the temporal-context choice is dimension-dependent: a TSN-like bidirectional LSTM over four clips helps valence, while an exponential moving average with decay weight 0.98 helps arousal by smoothing its slow-changing intensity. The authors present these as integrated findings, not as separate tricks, since each design choice is motivated by an observed property of emotion dynamics.

Load-bearing premise

The results assume that choosing the arousal decay weight, the valence clip count, and each task's modality set by looking at test-set outcomes does not inflate the reported scores, since no validation-set experiment is shown to confirm those choices generalize.

Editorial extensions

If this is right

  • On the LIRIS-ACCEDE test set, the reported framework improves valence MSE and arousal PCC over every state-of-the-art system listed in the comparison table, including the strongest prior GLA baseline.
  • Audio features alone carry the most emotional signal, so any successor system that drops or weakens the audio stream is likely to lose performance on this task.
  • Because the progressive training makes per-clip modality selection dynamic, the fused model can lean on whatever modality is discriminative for a given clip, which is a mechanism the paper argues that concat-based or voting-based fusion lacks.
  • The valence long-context result of four clips suggests there is an optimal context window for valence prediction, with shorter windows losing context and longer windows adding noise from abrupt emotional shifts.
  • The arousal result shows that a parameter-free exponential moving average can beat more complex recurrent temporal modeling for arousal, supporting the paper's premise that arousal changes slowly and its trend is strongly auto-regressive.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the reported margin holds under a proper held-out validation split, the same recipe—per-task modality selection plus residual fusion—could transfer to other continuous affect datasets, but the specific rankings (audio first, expression only for valence) may shift with genre or culture.
  • The exponential moving average result is an implicit argument that arousal can be modeled with a low-order autoregressive process; a natural testable extension is replacing the fixed beta with a learned gating mechanism or comparing against a Kalman filter.
  • The residual progressive training can be read as a general way to combine modalities of unequal quality, so it may be worth trying on other multi-modal prediction problems, not just emotion, whenever feature-level fusion tends to overfit.
  • The paper's bad-case analysis suggests that comedy clips with fighting or quarrels are systematically mispredicted as negative valence; a testable extension would add genre-aware features or some form of situational/narrative context to correct this.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes a multi-modal framework for continuous valence/arousal prediction on the LIRIS-ACCEDE dataset. Features are extracted with pre-trained audio (VGGish), facial expression, scene, and action (OpenPose) modules; per-modality BiLSTMs provide intra-clip temporal fusion, while inter-clip context is modeled with a TSN-like BiLSTM for valence and an exponential moving average for arousal. The model is trained with a residual-based progressive strategy, and the authors report comparisons to prior MediaEval 2018 systems as well as MediaEval 2015 classification results. The central claim is that the proposed framework achieves better performance than the state of the art by a large margin.

Significance. If the claims were supported, the paper would be a useful contribution to affective video analysis: it combines modality-specific deep features, an interpretable sequential fusion procedure, and separate temporal treatments for valence and arousal. The manuscript also deserves credit for reporting per-modality ablations and for comparing with several published systems on the same dataset. However, the evaluation as reported does not establish the headline claim: hyperparameters and per-task modality sets are selected using test-set performance, and the arousal MSE in the paper's own comparison table is worse than the best competitor under the official metric. The reported large-margin advantage is therefore not credible, and the framework's significance relative to prior art remains unclear.

major comments (4)
  1. [Experiments and Results, Performance for Long Temporal Fusion (Tables 3 and 4)] The final valence and arousal results are obtained by selecting hyperparameters on the test set: the number of context clips is chosen as 4 and the EMA decay beta as 0.98 (or 0.99) because these values give the best numbers in Tables 3 and 4. No development split is used, so the reported valence MSE of 0.071 and arousal PCC of 0.419 are optimistically biased by test-set selection. In addition, Table 4 is internally ambiguous: beta=0.96/0.97 gives the best MSE (0.136), beta=0.99 gives the best PCC (0.427), and the reported arousal MSE of 0.137 corresponds to beta=0.98 without a stated selection criterion. The authors should select these hyperparameters on a validation split and then report test results once.
  2. [Modality-specific Performance (Table 1) and Multi-modal Features] The per-task modality sets are chosen after evaluating modality-specific models on the test set: Table 1 ranks audio, scene, expression, and action on the test set, and the text then excludes action for valence and expression for arousal. Because the same test set is used for the final comparison in Table 5, this modality selection is circular. Modality selection must be performed on a development subset, or with nested cross-validation, and only the final evaluation should touch the test set.
  3. [Comparison with The State-of-the-art (Table 5)] The abstract's claim of 'better performance ... with a large margin' is contradicted by the paper's own table under the official metric: the proposed method's arousal MSE (0.137) is worse than GLA's (0.133). The large margin exists only for valence MSE/PCC. The claim should be restricted to valence, or the arousal model needs to be modified and re-evaluated; as written, the conclusion overstates the results.
  4. [Experiments and Results] The experimental section reports single-run numbers on a 12-movie test set with no error bars, per-movie breakdown, or significance testing. Differences as small as 0.001-0.01 in MSE are quoted as decisive advantages, and the PCC differences are not assessed for statistical reliability. The authors should report results across runs, movies, or bootstrap intervals before claiming superiority.
minor comments (5)
  1. [Abstract] The abstract contains typos such as 'temporal contex' and 'audiences emotional states'; these should be corrected.
  2. [Dataset and Metrics] The paper states that MSE is the official metric, but the hyperparameter choices in Tables 3 and 4 appear to optimize different quantities for valence and arousal; the selection criterion should be stated explicitly, since selecting beta by PCC while reporting MSE is inconsistent.
  3. [Results, Comparison with MediaEval 2015] Table 6 reports classification accuracies on the MediaEval 2015 task, which uses different evaluation protocols and annotations; these results cannot be used as evidence for the valence/arousal regression framework, and this comparison should be clearly separated from the main regression evaluation.
  4. [Figures 3 and 4] The T-SNE plots are described as demonstrating discriminability, but no quantitative measure is given; consider replacing or supplementing them with a quantitative separability metric.
  5. [Equation (2)] The EMA is defined on clip-level predictions, but the per-second evaluation repeats each clip prediction 10 times; the interaction between the EMA and this repetition should be clarified.

Circularity Check

3 steps flagged · score 6.0 of 10

The headline valence/arousal numbers are the best values of a grid evaluated on the test set, and modality sets are chosen from the same Results section, so the 'large-margin' claim is partly a test-set selection artifact.

  1. fitted input called prediction [Experiments and Results, 'Performance for Long Temporal Fusion', Tables 3 and 5]
    "We try different clips number for the valence task and different decay weights β for the arousal task to get the best performance. ... Table 3: number of clips 4, MSE 0.071, PCC 0.444. ... Table 5: Ours 0.071 0.444 0.137 0.419."

    The valence row reported as 'Ours' in Table 5 is exactly the best row of Table 3 (4 clips, MSE 0.071, PCC 0.444). The paper states that the clip count was chosen by trying different values 'to get the best performance,' and it reports no validation split between this selection and the test set. Thus the headline valence result is not an independent prediction of a fixed model; it is the selected best of a test-set grid search, so the reported value is partly constructed from the same data on which it is evaluated.

  2. fitted input called prediction [Experiments and Results, 'Performance for Long Temporal Fusion', Tables 4 and 5]
    "We try different clips number for the valence task and different decay weights β for the arousal task to get the best performance. ... Table 4: decay weights β ... 0.98 0.137 0.419. ... Table 5: Ours 0.071 0.444 0.137 0.419."

    The arousal row reported as 'Ours' in Table 5 (MSE 0.137, PCC 0.419) is one of the beta values tried in Table 4 (β=0.98). The decay weight is selected from test-set performance, with no validation split, so the final arousal evaluation is not on an untouched test set. The reported arousal MSE/PCC is the outcome of a test-set hyperparameter search rather than a prediction of a model whose hyperparameters were fixed before seeing the test set.

1 more flagged steps
  1. other [Experiments and Results, 'Modality-specific Performance' and 'Performance for Intra-clip Part', Table 1 and Figure 3]
    "To evaluate the importance of every modality for each task, we train modality-specific models. The performance of each model is shown in Table. 1. Besides, to better demonstrate each modality's effectiveness, we visualize the features extracted from clips in the test set by the modality-specific models. ... According to the analysis and experiments, audio and scenes are used for both valence and arousal tasks while expressions are only used in the valence task and actions are only used in the arousal task."

    The final per-task modality sets are decided after the modality-specific performance table and the test-set feature visualization are inspected, and no validation split is reported separating this choice from the test set. The final multimodal model is then evaluated on the same test set, so the reported valence and arousal numbers inherit the modality-selection signal. This is model selection on the evaluation data rather than an independent prediction of a pre-specified architecture.

full rationale

The model equations (MSE losses, EMA in Eq. 2, residual mapping H_i(x)=f_i-f_{i-1}) are not circular: no derived quantity is defined in terms of the target it predicts. The circularity is in the evaluation protocol. The paper reports no validation split, and the final 'Ours' rows in Table 5 coincide with hyperparameter values that the text says were chosen by trying values 'to get the best performance' (Tables 3 and 4), while the modality sets are chosen from the same Results section (Table 1/Figure 3). Thus the headline valence and arousal numbers are partly selected on the test set rather than independently predicted. Separately, under the official MSE metric, the arousal MSE 0.137 in Table 5 is not better than GLA's 0.133; that is a correctness concern, not a circularity step, and does not change this score.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim depends on two test-set-selected hyperparameters and test-set-selected modality sets, plus several untested psychological and transfer assumptions. No new entities are invented.

free parameters (3)
  • EMA decay beta = 0.98
    Selected by comparing test-set MSE/PCC over beta in {0.96, 0.97, 0.98, 0.99}; beta=0.98 gives the reported tradeoff.
  • Valence context clip count = 4
    Selected by comparing test-set MSE/PCC over clip counts {3, 4, 5, 6}; clip count 4 gives best valence result.
  • Per-task modality sets = valence: audio+scene+expression; arousal: audio+scene+action
    Chosen after evaluating each modality alone on the test set (Table 1); this is a data-driven selection on the evaluation set.
assumptions (3)
  • domain assumption Valence is better captured by facial expressions, while arousal is better captured by body actions.
    Stated in 'Multi-modal Features' with examples; used to decide which modalities to include for each task.
  • domain assumption Arousal changes slowly over short time, so an exponential moving average is appropriate; valence is more context-dependent and requires LSTM.
    Invoked in 'Inter-clip Long Temporal Fusion' to justify different temporal models for valence and arousal.
  • domain assumption Pre-trained models trained on ImageNet, RAF, Places365, and pose data transfer their features to movie emotion prediction.
    The paper freezes feature extractors and uses them on the LIRIS-ACCEDE domain; this transferability is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Video Affective Effects Prediction with Multi-modal Fusion and Shot-Long Temporal Context." pith.science (2026). https://pith.science/paper/IUXU5D7M

@misc{pith2026190901763,
  author       = {Pith},
  title        = {Pith review of: Video Affective Effects Prediction with Multi-modal Fusion and Shot-Long Temporal Context},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IUXU5D7M}},
  note         = {Machine review of arXiv:1909.01763}
}
read the original abstract

Predicting the emotional impact of videos using machine learning is a challenging task considering the varieties of modalities, the complicated temporal contex of the video as well as the time dependency of the emotional states. Feature extraction, multi-modal fusion and temporal context fusion are crucial stages for predicting valence and arousal values in the emotional impact, but have not been successfully exploited. In this paper, we propose a comprehensive framework with novel designs of modal structure and multi-modal fusion strategy. We select the most suitable modalities for valence and arousal tasks respectively and each modal feature is extracted using the modality-specific pre-trained deep model on large generic dataset. Two-time-scale structures, one for the intra-clip and the other for the inter-clip, are proposed to capture the temporal dependency of video content and emotion states. To combine the complementary information from multiple modalities, an effective and efficient residual-based progressive training strategy is proposed. Each modality is step-wisely combined into the multi-modal model, responsible for completing the missing parts of features. With all those improvements above, our proposed prediction framework achieves better performance on the LIRIS-ACCEDE dataset with a large margin compared to the state-of-the-art.

Figures

Figures reproduced from arXiv: 1909.01763 by the authors.

Figure 1
Figure 1. Part (a) describes our overall framework for valence and arousal tasks separately. Our framework involves two parts: [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Residual-based progressive training strategy for intra-clip feature fusion of valence [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 4
Figure 4. Features extracted from movie clips in the test set [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figures from the paper (3 more)
Figure 6
Figure 6. Figure 6: The predictions of movies in the test set for the [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Bad cases in the test set. The left one is the valence [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 5
Figure 5. Figure 5: The predictions of movies in test set for the valence [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 40 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTIO...

  2. [2]

    K.; Hossain, M

    Atrey, P. K.; Hossain, M. A.; El Saddik, A.; and Kankanhalli, M. S. 2010. Multimodal fusion for multimedia analysis: a survey. Multimedia systems 16(6):345--379

    work page 2010

  3. [3]

    F.; Lindquist, K

    Barrett, L. F.; Lindquist, K. A.; Bliss-Moreau, E.; Duncan, S.; Gendron, M.; Mize, J.; and Brennan, L. 2007. Of mice and men: Natural kinds of emotions in the mammalian brain? a response to panksepp and izard. Perspectives on Psychological Science 2(3):297--312

    work page 2007

  4. [4]

    Batziou, E.; Michail, E.; Avgerinakis, K.; Vrochidis, S.; Patras, I.; and Kompatsiaris, I. 2018. Visual and audio analysis of movies video for emotion detection @ emotional impact of movies task mediaeval 2018. In MediaEval

    work page 2018

  5. [5]

    Cao, Z.; Hidalgo, G.; Simon, T.; Wei, S.-E.; and Sheikh, Y. 2018. Open P ose: realtime multi-person 2 D pose estimation using P art A ffinity F ields. In arXiv preprint arXiv:1812.08008

    arXiv 2018

  6. [6]

    K.; Pandharipande, M.; Hassan, E.; Ghosh, H.; and Kopparapu, S

    Chakraborty, R.; Maurya, A. K.; Pandharipande, M.; Hassan, E.; Ghosh, H.; and Kopparapu, S. K. 2015. Tcs-ilab-mediaeval 2015: Affective impact of movies and violent scene detection. In MediaEval

    work page 2015

  7. [7]

    Cho, K.; Van Merri \"e nboer, B.; Gulcehre, C.; Bahdanau, D.; Bougares, F.; Schwenk, H.; and Bengio, Y. 2014. Learning phrase representations using rnn encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078

    arXiv 2014

  8. [8]

    Chollet, F. 2017. Xception: Deep learning with depthwise separable convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition , 1251--1258

    work page 2017

Show all 44 references
  1. [9]

    Cortes, C., and Vapnik, V. 1995. Support-vector networks. Machine learning 20(3):273--297

    1995

  2. [10]

    Cowie, R.; Douglas-Cowie, E.; Tsapatsoulis, N.; Votsis, G.; Kollias, S.; Fellenz, W.; and Taylor, J. G. 2001. Emotion recognition in human-computer interaction. IEEE Signal processing magazine 18(1):32--80

    2001

  3. [11]

    Dai, Q.; Zhao, R.-W.; Wu, Z.; Wang, X.; Gu, Z.; Wu, W.; and Jiang, Y.-G. 2015. Fudan-huawei at mediaeval 2015: Detecting violent scenes and affective impact in movies with deep learning. In MediaEval

    2015

  4. [12]

    Deng, J.; Dong, W.; Socher, R.; Li, L.-J.; Li, K.; and Fei-Fei, L. 2009. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition , 248--255. Ieee

    2009

  5. [13]

    H., and Reeves, B

    Detenber, B. H., and Reeves, B. 1996. A bio-informational theory of emotion: Motion and image size effects on viewers. Journal of Communication

    1996

  6. [14]

    Dobri s ek, S.; Gaj s ek, R.; Miheli c , F.; Pave s i \'c , N.; and S truc, V. 2013. Towards efficient multi-modal emotion recognition. International Journal of Advanced Robotic Systems 10(1):53

    2013

  7. [15]

    Fan, Y.; Lu, X.; Li, D.; and Liu, Y. 2016. Video-based emotion recognition using cnn-rnn and c3d hybrid networks. In Proceedings of the 18th ACM International Conference on Multimodal Interaction , 445--450. ACM

    2016

  8. [16]

    Gan, Q.; Wang, S.; Hao, L.; and Ji, Q. 2017. A multimodal deep regression bayesian network for affective video content analyses. In Proceedings of the IEEE International Conference on Computer Vision , 5113--5122

    2017

  9. [17]

    P.; Gemmeke, J

    Hershey, S.; Chaudhuri, S.; Ellis, D. P.; Gemmeke, J. F.; Jansen, A.; Moore, R. C.; Plakal, M.; Platt, D.; Saurous, R. A.; Seybold, B.; et al. 2017. Cnn architectures for large-scale audio classification. In 2017 ieee international conference on acoustics, speech and signal pr...

    2017

  10. [18]

    Hochreiter, S., and Schmidhuber, J. 1997. Long short-term memory. Neural computation 9(8):1735--1780

    1997

  11. [19]

    Izard, C. E. 2007. Basic emotions, natural kinds, emotion schemas, and a new paradigm. Perspectives on psychological science 2(3):260--280

    2007

  12. [20]

    H.; Gu, Z.; He, T.; and Liu, Y

    Ko, T. H.; Gu, Z.; He, T.; and Liu, Y. 2018. Towards learning emotional subspace. In MediaEval

    2018

  13. [21]

    Kurpukdee, N.; Koriyama, T.; Kobayashi, T.; Kasuriya, S.; Wutiwiwatchai, C.; and Lamsrichan, P. 2017. Speech emotion recognition using convolutional long short-term memory neural network and support vector machines. In 2017 Asia-Pacific Signal and Information Processing Associ...

    2017

  14. [22]

    Li, S.; Deng, W.; and Du, J. 2017. Reliable crowdsourcing and deep locality-preserving learning for expression recognition in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition , 2852--2861

    2017

  15. [23]

    Liu, C.; Tang, T.; Lv, K.; and Wang, M. 2018. Multi-feature based emotion recognition for video clips. In Proceedings of the 2018 on International Conference on Multimodal Interaction , 630--634. ACM

    2018

  16. [24]

    Ma, Y.; Liang, X.; and Xu, M. 2018. Thuhcsi in mediaeval 2018 emotional impact of movies task. In MediaEval

    2018

  17. [25]

    Marin Vlastelica, P.; Hayrapetyan, S.; Tapaswi, M.; and Stiefelhagen, R. 2015. Kit at mediaeval 2015-evaluating visual cues for affective impact of movies task. In MediaEval

    2015

  18. [26]

    Metallinou, A.; Lee, S.; and Narayanan, S. 2010. Decision level combination of multiple modalities for recognition and analysis of emotional expression. In 2010 IEEE International Conference on Acoustics, Speech and Signal Processing , 2462--2465. IEEE

    2010

  19. [27]

    Mironica, I.; Ionescu, B.; Sj \"o berg, M.; Schedl, M.; and Skowron, M. 2015. Rfa at mediaeval 2015 affective impact of movies task: A multimodal approach. In MediaEval

    2015

  20. [28]

    Moreira, D.; de Avila, S. E. F.; Perez, M.; Moraes, D.; Testoni, V.; Valle, E.; Goldenstein, S.; and Rocha, A. 2015. Recod at mediaeval 2015: Affective impact of movies task. In MediaEval

    2015

  21. [29]

    Pang, L.; Zhu, S.; and Ngo, C.-W. 2015. Deep multimodal learning for affective analysis and retrieval. IEEE Transactions on Multimedia 17(11):2008--2020

    2015

  22. [30]

    C.; Nguyen, V.-T.; and Tran, M.-T

    Quan, K.-A. C.; Nguyen, V.-T.; and Tran, M.-T. 2018. Frame-based evaluation with deep features to predict emotional impact of movies. In MediaEval

    2018

  23. [31]

    P.; Mihalcea, R.; and Morency, L.-P

    Rosas, V. P.; Mihalcea, R.; and Morency, L.-P. 2013. Multimodal sentiment analysis of spanish online videos. IEEE Intelligent Systems 28(3):38--45

    2013

  24. [32]

    Seddati, O.; Kulah, E.; Pironkov, G.; Dupont, S.; Mahmoudi, S.; and Dutoit, T. 2015. Umons at mediaeval 2015 affective impact of movies task including violent scenes detection. In MediaEval

    2015

  25. [33]

    L.; Ionescu, B.; Dellandr \'e a, E.; Schedl, M.; Demarty, C.-H.; and Chen, L

    Sj \"o berg, M.; Baveye, Y.; Wang, H.; Quang, V. L.; Ionescu, B.; Dellandr \'e a, E.; Schedl, M.; Demarty, C.-H.; and Chen, L. 2015. The mediaeval 2015 affective impact of movies task. In MediaEval

    2015

  26. [34]

    J.; Liu, T.; and Prasad, G

    Sun, J. J.; Liu, T.; and Prasad, G. 2018. Gla in mediaeval 2018 emotional impact of movies task. In MediaEval

    2018

  27. [35]

    Tran, D.; Bourdev, L.; Fergus, R.; Torresani, L.; and Paluri, M. 2015. Learning spatiotemporal features with 3d convolutional networks. In Proceedings of the IEEE international conference on computer vision , 4489--4497

    2015

  28. [36]

    affective impact of movies

    Trigeorgis, G.; Coutinho, E.; Ringeval, F.; Marchi, E.; Zafeiriou, S.; and Schuller, B. W. 2015. The icl-tum-passau approach for the mediaeval 2015" affective impact of movies" task. In MediaEval

    2015

  29. [37]

    Vu Lam, S. P. L., and Le, D.-D. Shin’ichi satoh, and duc anh duong. 2015. nii-uit at mediaeval 2015 affective impact of movies task. In Working Notes Proceedings of the MediaEval 2015 Workshop, Wurzen, Germany, September , 14--15

    2015

  30. [38]

    L., and Cheong, L.-F

    Wang, H. L., and Cheong, L.-F. 2006. Affective understanding in film. IEEE Transactions on circuits and systems for video technology 16(6):689--704

    2006

  31. [39]

    Wang, L.; Xiong, Y.; Wang, Z.; Qiao, Y.; Lin, D.; Tang, X.; and Van Gool, L. 2016. Temporal segment networks: Towards good practices for deep action recognition. In European conference on computer vision , 20--36. Springer

    2016

  32. [40]

    Wimmer, M.; Schuller, B.; Arsic, D.; Radig, B.; and Rigoll, G. 2008. Low-level fusion of audio and video feature for multi-modal emotion recognition. In Proc. 3rd Int. Conf. on Computer Vision Theory and Applications VISAPP, Funchal, Madeira, Portugal , 145--151

    2008

  33. [41]

    S.; Luo, S.; and Rui, Y

    Xu, M.; Xu, C.; He, X.; Jin, J. S.; Luo, S.; and Rui, Y. 2013. Hierarchical affective content analysis in arousal and valence dimensions. Signal Processing 93(8):2140--2150

    2013

  34. [42]

    Yi, Y.; Wang, H.; Zhang, B.; and Yu, J. 2015. Mic-tju in mediaeval 2015 affective impact of movies task. In MediaEval

    2015

  35. [43]

    Yi, Y.; Wang, H.; and Li, Q. 2018. Cnn features for emotional impact of movies task. In MediaEval

    2018

  36. [44]

    Zhang, K.; Zhang, Z.; Li, Z.; and Qiao, Y. 2016. Joint face detection and alignment using multitask cascaded convolutional networks. IEEE Signal Processing Letters 23(10):1499--1503

    2016

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.