Seeing the Arrow of Time

By L. Pickup, Zheng Pan, Donglai Wei, Yichang Shih, Changshui Zhang, Andrew Zisserman, Bernhard Schölkopf, William T. Freeman. 2014

111 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Generating Videos with Scene Dynamics (2016, background): “Conceptually, our work is related to studies into fundamental roles of time in computer vision [30, 12, 2, 7, 24].”
Visual7W: Grounded Question Answering in Images (2015, background): “The 7W questions roughly correspond to an array of standard vision tasks: what [9, 15, 39], where [24, 50], when [30, 32], who [35, 42], why [33], how [23, 31] and which [16, 17].”
Rethinking Spatiotemporal Feature Learning: Speed-Accuracy Trade-offs in Video Classification (2017, background): “In the former case, the performance of I3D is indifferent to the “arrow of time” [37], whereas in the latter case, reversing the order hurts performance.”
UvA-DARE (Digital Academic Repository) Self-Supervised Video Representation Learning With Odd-One-Out Networks (neutral): “Besides, unlabeled videos are abundant and information rich regarding spatio-temporal structures [5, 33, 3, 32, 45].”
Discovering Causal Signals in Images (2016, neutral): “[24] use observational causal discovery techniques to determine the direction of time in video playback.”
The Arrow of Time in Multivariate Time Series (2016, methodology): “In a computer vision application, Pickup et al. (2014) aim to determine if videos are shown in forward or backward direction.”
Algorithmic independence of initial condition and dynamical law in thermodynamics and causal inference (2015, background): “in finance or brain research) or to infer the time direction of movies [16].”
Learning and Using the Arrow of Time (2018, methodology): “On the previous benchmark TA-180 [14], we only test with models trained on Flickr-AoT or Kinetics-AoT dataset, as the dataset is too small to train our model.”
Don't just listen, use your imagination: Leveraging visual common sense for non-visual tasks (2015, background): “Example tasks include reasoning about what people talk about in images [4], understanding the flow of time (when) [39], identifying where the image is taken [24, 28] and judging the intentions of people in images (why) [40].”
Anticipating the future by watching unlabeled video (2015, background): “in [28] implicitly model causality to understand what should happen before what in a video.”
Anticipating Visual Representations from Unlabeled Video (2015, background): “Pickup et al. in [27] implicitly model causality to understand what should happen before what in a video.”
Object Level Visual Reasoning in Videos (2018, methodology): “Our approach is related to the concept of the “arrow of the time” [28] involving the “one-way direction” or “asymmetry” of time.”
Temporal Perception and Prediction in Ego-Centric Video (2015, methodology): “For specific information and parameter settings, please refer to [29].”
Learning Common Sense through Visual Abstraction (2015, background): “Recently, there has been a surge in interest in high-level “beyond recognition” tasks which can beneﬁt from external knowledge beyond what is depicted in the image [3, 19, 23, 30, 31].”
A Survey of Learning Causality with Data: Problems and Methods (2018, background): “As causality can be implied by the arrow of time [109], it can be defined in various ways for time series data.”
Evolving Losses for Unsupervised Video Representation Learning (2020, methodology): “Some tasks take advantage of the temporal structure in videos, such as predicting if frames appear in order, reverse order, shufﬂed, color-consistency across frames, etc. [12, 26, 30, 24, 31, 46, 19, 21, 44, 45, 41].”
BABEL: Bodies, Action and Behavior with English Labels (2021, unknown): “Unidirectional transitions such as sit → stand up and walk → sit implicitly indicate the arrow of time [24].”
A Review on Deep Learning Techniques for Video Prediction (2020, unknown): “literature as a supervisory signal [35]–[37].”
Evolving Losses for Unlabeled Video Representation Learning (2019, background): “Related Works There are many tasks for self-supervised learning such as predicting if frames appear in order [6, 5, 7], reconstruction or prediction of future frames [9], timecontrastive learning [3, 8].”
Rethinking Spatiotemporal Feature Learning For Video Understanding (2017, background): “We also see that I3D’s performance on the randomly shuffled data is much worse than on the normal form of the data; however, its performance on the reversed form is the same, indicating that this model and/or dataset does not allow (or require) inferring the causal “arrow of time” [33].”