arXiv subjects

Rui Dai

Publications and source records attributed to Rui Dai.

45 records · Page 3Linked to original sources

arXiv2021

Learning an Augmented RGB Representation with Cross-Modal Knowledge Distillation for Action Detection

In video understanding, most cross-modal knowledge distillation (KD) methods are tailored for classification tasks, focusing on the discriminative representation of the trimmed videos. However, action detection requires not only categorizing actions, but also localizing them in untrimmed videos. Therefore, transferring knowledge pertaining to temporal relations is critical for this task which is missing in the previous cross-modal KD frameworks. To this end, we aim at learning an augmented RGB representation for action detection, taking advantage of additional modalities at training time through KD. We propose a KD framework consisting of two levels of distillation. On one hand, atomic-level distillation encourages the RGB student to learn the sub-representation of the actions from the teacher in a contrastive manner. On the other hand, sequence-level distillation encourages the student to learn the temporal knowledge from the teacher, which consists of transferring the Global Contextual Relations and the Action Boundary Saliency. The result is an Augmented-RGB stream that can achieve competitive performance as the two-stream network while using only RGB at inference time. Extensive experimental analysis shows that our proposed distillation framework is generic and outperforms other popular cross-modal distillation methods in action detection task.

cs.CV↗

arXiv2021

VPN++: Rethinking Video-Pose embeddings for understanding Activities of Daily Living

Many attempts have been made towards combining RGB and 3D poses for the recognition of Activities of Daily Living (ADL). ADL may look very similar and often necessitate to model fine-grained details to distinguish them. Because the recent 3D ConvNets are too rigid to capture the subtle visual patterns across an action, this research direction is dominated by methods combining RGB and 3D Poses. But the cost of computing 3D poses from RGB stream is high in the absence of appropriate sensors. This limits the usage of aforementioned approaches in real-world applications requiring low latency. Then, how to best take advantage of 3D Poses for recognizing ADL? To this end, we propose an extension of a pose driven attention mechanism: Video-Pose Network (VPN), exploring two distinct directions. One is to transfer the Pose knowledge into RGB through a feature-level distillation and the other towards mimicking pose driven attention through an attention-level distillation. Finally, these two approaches are integrated into a single model, we call VPN++. We show that VPN++ is not only effective but also provides a high speed up and high resilience to noisy Poses. VPN++, with or without 3D Poses, outperforms the representative baselines on 4 public datasets. Code is available at https://github.com/srijandas07/vpnplusplus.

cs.CV↗

arXiv2021

Mechanical Properties of Gradient Copper Nano-Gyroid Cellular Structures: A Molecular Dynamics Study

Advanced manufacturing (AM) technologies, such as nanoscale additive manufacturing process, enable the fabrication of nanoscale architected materials which has received great attention due to their prominent properties. However, few studies delve into the functional gradient cellular architecture on nanoscale. This work studied the gradient nano-Gyroid architected material made of copper (Cu) by molecular dynamic (MD) simulations. The result reveals that, unlike homogeneous architecture, gradient Gyroid not only shows novel layer-by-layer deformation behavior, but also processes significantly better energy absorption ability. Moreover, this deformation behavior and energy absorption are predictable and designable, which demonstrates its highly programmable potential.

physics.app-ph↗

arXiv2020

Selective Spatio-Temporal Aggregation Based Pose Refinement System: Towards Understanding Human Activities in Real-World Videos

Taking advantage of human pose data for understanding human activities has attracted much attention these days. However, state-of-the-art pose estimators struggle in obtaining high-quality 2D or 3D pose data due to occlusion, truncation and low-resolution in real-world un-annotated videos. Hence, in this work, we propose 1) a Selective Spatio-Temporal Aggregation mechanism, named SST-A, that refines and smooths the keypoint locations extracted by multiple expert pose estimators, 2) an effective weakly-supervised self-training framework which leverages the aggregated poses as pseudo ground-truth instead of handcrafted annotations for real-world pose estimation. Extensive experiments are conducted for evaluating not only the upstream pose refinement but also the downstream action recognition performance on four datasets, Toyota Smarthome, NTU-RGB+D, Charades, and Kinetics-50. We demonstrate that the skeleton data refined by our Pose-Refinement system (SSTA-PRS) is effective at boosting various existing action recognition models, which achieves competitive or state-of-the-art performance.

cs.CV↗

arXiv2020

VPN: Learning Video-Pose Embedding for Activities of Daily Living

In this paper, we focus on the spatio-temporal aspect of recognizing Activities of Daily Living (ADL). ADL have two specific properties (i) subtle spatio-temporal patterns and (ii) similar visual patterns varying with time. Therefore, ADL may look very similar and often necessitate to look at their fine-grained details to distinguish them. Because the recent spatio-temporal 3D ConvNets are too rigid to capture the subtle visual patterns across an action, we propose a novel Video-Pose Network: VPN. The 2 key components of this VPN are a spatial embedding and an attention network. The spatial embedding projects the 3D poses and RGB cues in a common semantic space. This enables the action recognition framework to learn better spatio-temporal features exploiting both modalities. In order to discriminate similar actions, the attention network provides two functionalities - (i) an end-to-end learnable pose backbone exploiting the topology of human body, and (ii) a coupler to provide joint spatio-temporal attention weights across a video. Experiments show that VPN outperforms the state-of-the-art results for action classification on a large scale human activity dataset: NTU-RGB+D 120, its subset NTU-RGB+D 60, a real-world challenging human activity dataset: Toyota Smarthome and a small scale human-object interaction dataset Northwestern UCLA.

cs.CV↗

arXiv2020

Hybrid Spatio-Temporal Graph Convolutional Network: Improving Traffic Prediction with Navigation Data

Traffic forecasting has recently attracted increasing interest due to the popularity of online navigation services, ridesharing and smart city projects. Owing to the non-stationary nature of road traffic, forecasting accuracy is fundamentally limited by the lack of contextual information. To address this issue, we propose the Hybrid Spatio-Temporal Graph Convolutional Network (H-STGCN), which is able to "deduce" future travel time by exploiting the data of upcoming traffic volume. Specifically, we propose an algorithm to acquire the upcoming traffic volume from an online navigation engine. Taking advantage of the piecewise-linear flow-density relationship, a novel transformer structure converts the upcoming volume into its equivalent in travel time. We combine this signal with the commonly-utilized travel-time signal, and then apply graph convolution to capture the spatial dependency. Particularly, we construct a compound adjacency matrix which reflects the innate traffic proximity. We conduct extensive experiments on real-world datasets. The results show that H-STGCN remarkably outperforms state-of-the-art methods in various metrics, especially for the prediction of non-recurring congestion.

cs.LG↗

arXiv2020

Testing kinetically coupled inflation models with CMB distortions

Inflation scenarios kinetically coupled with the Einstein tensor have been widely studied. They can be consistent with current observational data. Future experiments on the measurement on CMB distortions will potentially extend information about the scalar spectrum to small scales $1 \Mpc^{-1} \lesssim k \lesssim 2 \times 10^4 \Mpc^{-1}$. By taking the sensitivity of the PIXIE experiment as the criterion, we perform a model-oriented analysis of the observational prospects of spectral distortions for kinetically coupled inflation. There are five models that possibly generate a detectable level of distortions, among the 49 single-field inflation models listed in Ref. \cite{Martin2013a}. These models are: hybrid inflation in the valley (VHI), non-canonical Kähler inflation (NCKI), generalized MSSM inflation (GMSSMI), generalized renormalization point inflation (GRIPI), and running-mass inflation (RMI). Each of these models can satisfy the Planck constraints on spectral tilt and lead to increase power on scales relevant for CMB distortions in a tuned region of their parameter space. The existence of kinetic coupling suppresses the value of the model parameters with mass dimension for VHI, GMSSMI, and GRIPI, such that these three models can be in agreement with their theoretical considerations. However, the tuned regions for all these models fail to satisfy the constraints on tensor modes.

astro-ph.CO↗

arXiv2018

Convolutional Gated Recurrent Units for Medical Relation Classification

Convolutional neural network (CNN) and recurrent neural network (RNN) models have become the mainstream methods for relation classification. We propose a unified architecture, which exploits the advantages of CNN and RNN simultaneously, to identify medical relations in clinical records, with only word embedding features. Our model learns phrase-level features through a CNN layer, and these feature representations are directly fed into a bidirectional gated recurrent unit (GRU) layer to capture long-term feature dependencies. We evaluate our model on two clinical datasets, and experiments demonstrate that our model performs significantly better than previous single-model methods on both datasets.

cs.CL↗

arXiv2018

Classifying medical relations in clinical text via convolutional neural networks

Deep learning research on relation classification has achieved solid performance in the general domain. This study proposes a convolutional neural network (CNN) architecture with a multi-pooling operation for medical relation classification on clinical records and explores a loss function with a category-level constraint matrix. Experiments using the 2010 i2b2/VA relation corpus demonstrate these models, which do not depend on any external features, outperform previous single-model methods and our best model is competitive with the existing ensemble-based method.

cs.CL↗