120 fps test的問題，透過圖書和論文來找解法和答案更準確安心。 我們找到下列懶人包和總整理

120 fps test進入發燒排行的影片

通過時間依賴性建模了解視頻上下文

為了解決120 fps test的問題，作者狄騠克 這樣論述：

Video context understanding has attracted increasing interests due to its potential applications in a wide range of areas. However, analyzing context within videos is not a straightforward task due to a number of factors like camera movement, multiple view- points, low-resolution quality, illuminat

ion, occlusion, and inter-class variation. Mean- while, learning temporal dependency has been demonstrated to be beneficial for video understanding as videos contain not only spatial but also temporal information. Thus, this dissertation aims to develop a number of algorithms to effectively recogniz

e human behaviors in a variety of different scenarios by leveraging temporal dependency across the video frames. We focus on three difficult, yet important tasks: first-person action recognition, extreme low-resolution action recognition, and anomaly detection.First, we present a framework for first

-person action recognition with a combination of temporal pooling and the Hilbert–Huang transform (HHT). It first adaptively performs temporal sub-action localization, treats each channel of the extracted trajectory pooled convolutional neural network (CNN) features as a time series, and summarizes

the temporal dynamic information in each sub-action by temporal pooling. The temporal evolution across sub-actions is then modeled by rank pooling. Thereafter, to account for the highly dynamic scene changes in first-person videos, the HHT is employed to de- compose the ranked pooling features into

finite and often few data-dependent functions, called intrinsic mode functions (IMFs), through empirical mode decomposition. Hilbert spectral analysis is then applied to each IMF component, and four salient descriptors are scrutinized and aggregated into the final video descriptor. Such a framework

cannot only precisely acquire both long- and short-term tendencies, but also address the cumbersome significant camera motion in first-person videos to render better accuracy.Second, we present a novel three-stream network for action recognition in extreme low resolution (LR) videos. In contrast to

the existing networks, the new network uses the trajectory-spatial network, which is robust against visual distortion, instead of the pose information to complement the two-stream network. Also, the new three-stream network is combined with the inflated 3D ConvNet (I3D) model pre-trained on kinetics

to produce more discriminative spatio-temporal features in blurred LR videos. Moreover, a bidirectional self-attention network is aggregated with the three-stream network to further manifest various temporal dependence among the spatio-temporal features. A new fusion strategy is devised as well to

integrate the information from the three different modalities.Third, we present a novel weakly supervised approach for anomaly detection, which begins with a relation-aware feature extractor to capture the multi-scale CNN features from a video. Afterwards, self-attention is integrated with condition

al random fields (CRFs), the core of the network, to make use of the ability of self-attention in capturing the short-range correlations of the features and the ability of CRFs in learning the inter-dependencies of these features. Such a framework can learn not only the dynamic interactions among th

e actors which are important for detecting complex movements, but also their short- and long-term dependencies across frames. Also, to deal with both local and non-local relationships of the features, a new variant of self-attention is developed by taking into consideration of a set of cliques with

different temporal localities. More- over, a new loss function which take the advantage of contrastive loss with multi-instance learning is considered to broaden the gap between the normal and abnormal samples, resulting in more accurate abnormal discrimination. Finally, the framework also extended

into an online setting, which enables real-time low-latency anomaly detection and ported on limited resources devices such as Jetson Nano.