Play Fair: Frame Attributions in Video Models

Price, Will; Damen, Dima

doi:10.1007/978-3-030-69541-5_29

Cited by 6 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Huang [23] study the effect of motion in action recognition by checking the accuracy drop from frames without using motion. Price et al [37] explores frame-level contribution to the model output with Element Shapley Value. It should be noted that most of prior works try to find salient spatio-temporal regions for action recognition while our work quantifies how a model learns temporal relevances and investigate inter-frame relationship based on its architecture, as opposed to why a model makes a specific prediction.…”

Section: Related Workmentioning

confidence: 99%

Temporal Relevance Analysis for Video Action Models

Fan¹,

Kim²,

Chun-Fu³

et al. 2022

Preprint

View full text Add to dashboard Cite

In this paper, we provide a deep analysis of temporal modeling for action recognition, an important but underexplored problem in the literature. We first propose a new approach to quantify the temporal relationships between frames captured by CNN-based action models based on layer-wise relevance propagation. We then conduct comprehensive experiments and in-depth analysis to provide a better understanding of how temporal modeling is affected by various factors such as dataset, network architecture, and input frames. With this, we further study some important questions for action recognition that lead to interesting findings. Our analysis shows that there is no strong correlation between temporal relevance and model performance; and action models tend to capture local temporal information, but less long-range dependencies. Our codes and models will be publicly available.

show abstract

Section: Related Workmentioning

confidence: 99%

Temporal Relevance Analysis for Video Action Models

Fan¹,

Kim²,

Chun-Fu³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Qualitative differences are found between what the two model types tend to use as evidence for classification decisions. Another work on explainability for video models is by Price et al [38], but only one type of model, and its decisions, is studied (TRN [54]). We are connected to the work of Sevilla-Lara et al [40], who discuss the risk that models with strong image modeling abilities may prioritize those cues over the temporal modeling cues.…”

Section: Related Workmentioning

confidence: 99%

Recur, Attend or Convolve? Frame Dependency Modeling Matters for Cross-Domain Robustness in Action Recognition

Broomé¹,

Pokropek²,

Li³

2021

Preprint

View full text Add to dashboard Cite

Most action recognition models today are highly parameterized, and evaluated on datasets with predominantly spatially distinct classes. Previous results for single images have shown that 2D Convolutional Neural Networks (CNNs) tend to be biased toward texture rather than shape for various computer vision tasks [16], reducing generalization. Taken together, this raises suspicion that large video models learn spurious correlations rather than to track relevant shapes over time and infer generalizable semantics from their movement. A natural way to avoid parameter explosion when learning visual patterns over time is to make use of recurrence across the time-axis. In this article, we empirically study the cross-domain robustness for recurrent, attention-based and convolutional video models, respectively, to investigate whether this robustness is influenced by the frame dependency modeling. Our novel Temporal Shape dataset is proposed as a light-weight dataset to assess the ability to generalize across temporal shapes which are not revealed from single frames. We find that when controlling for performance and layer structure, recurrent models show better out-of-domain generalization ability on the Temporal Shape dataset than convolution-and attention-based models. Moreover, our experiments indicate that convolution-and attention-based models exhibit more texture bias on Diving48 than recurrent models.

show abstract

“…Bargal et al [2] propose an explanation technique for recurrent neural networks (RNNs) with convolutional layers utilizing excitation backpropagation [27]. Perturbationbased black-box approaches have also been investigated to explain a video classifier by presenting salient frames [18] or a 3D generalization of a saliency map [14]. Same as the explanation of image classifiers, our technique allows combining the techniques above and the Bayesian optimization to balance various tradeoffs.…”

Section: Introductionmentioning

confidence: 99%

BOREx: Bayesian-Optimization--Based Refinement of Saliency Map for Image- and Video-Classification Models

Kikuchi¹,

Uchida²,

Waga³

et al. 2022

Preprint

View full text Add to dashboard Cite

Explaining a classification result produced by an image-and video-classification model is one of the important but challenging issues in computer vision. Many methods have been proposed for producing heat-map-based explanations for this purpose, including ones based on the white-box approach that uses the internal information of a model (e.g., LRP, Grad-CAM, and Grad-CAM++) and ones based on the black-box approach that does not use any internal information (e.g., LIME, SHAP, and RISE). We propose a new black-box method BOREx (Bayesian Optimization for Refinement of visual model Explanation) to refine a heat map produced by any method. Our observation is that a heat-map-based explanation can be seen as a prior for an explanation method based on Bayesian optimization. Based on this observation, BOREx conducts Gaussian process regression (GPR) to estimate the saliency of each pixel in a given image starting from the one produced by another explanation method. Our experiments statistically demonstrate that the refinement by BOREx improves low-quality heat maps for image-and video-classification results.

show abstract

Play Fair: Frame Attributions in Video Models

Cited by 6 publications

References 37 publications

Temporal Relevance Analysis for Video Action Models

Temporal Relevance Analysis for Video Action Models

Recur, Attend or Convolve? Frame Dependency Modeling Matters for Cross-Domain Robustness in Action Recognition

BOREx: Bayesian-Optimization--Based Refinement of Saliency Map for Image- and Video-Classification Models

Contact Info

Product

Resources

About