Learning Dynamical Human-Joint Affinity for 3D Pose Estimation in Videos

Graph Convolution Network (GCN) has been successfully used for 3D human pose estimation in videos. However, it is often built on the fixed human-joint affinity, according to human skeleton. This may reduce adaptation capacity of GCN to tackle complex spatio-temporal pose variations in videos. To alleviate this problem, we propose a novel Dynamical Graph Network (DG-Net), which can dynamically identify human-joint affinity, and estimate 3D pose by adaptively learning spatial/temporal joint relations from videos. Different from traditional graph convolution, we introduce Dynamical Spatial/Temporal Graph convolution (DSG/DTG) to discover spatial/temporal human-joint affinity for each video exemplar, depending on spatial distance/temporal movement similarity between human joints in this video. Hence, they can effectively understand which joints are spatially closer and/or have consistent motion, for reducing depth ambiguity and/or motion uncertainty when lifting 2D pose to 3D pose. We conduct extensive experiments on three popular benchmarks, e.g., Human3.6M, HumanEva-I, and MPI-INF-3DHP, where DG-Net outperforms a number of recent SOTA approaches with fewer input frames and model size.

Investigate Indistinguishable Points in Semantic Segmentation of 3D Point Cloud

This paper investigates the indistinguishable points (difficult to predict label) in semantic segmentation for large-scale 3D point clouds. The indistinguishable points consist of those located in complex boundary, points with similar local textures but different categories, and points in isolate small hard areas, which largely harm the performance of 3D semantic segmentation. To address this challenge, we propose a novel Indistinguishable Area Focalization Network (IAF-Net), which selects indistinguishable points adaptively by utilizing the hierarchical semantic features and enhances fine-grained features for points especially those indistinguishable points. We also introduce multi-stage loss to improve the feature representation in a progressive way. Moreover, in order to analyze the segmentation performances of indistinguishable areas, we propose a new evaluation metric called Indistinguishable Points Based Metric (IPBM). Our IAF-Net achieves the comparable results with state-of-the-art performance on several popular 3D point cloud datasets e.g. S3DIS and ScanNet, and clearly outperforms other methods on IPBM.

On the relationship between Gaussian stochastic blockmodels and label propagation algorithms

The problem of community detection receives great attention in recent years. Many methods have been proposed to discover communities in networks. In this paper, we propose a Gaussian stochastic blockmodel that uses Gaussian distributions to fit weight of edges in networks for non-overlapping community detection. The maximum likelihood estimation of this model has the same objective function as general label propagation with node preference. The node preference of a specific vertex turns out to be a value proportional to the intra-community eigenvector centrality (the corresponding entry in principal eigenvector of the adjacency matrix of the subgraph inside that vertexs community) under maximum likelihood estimation. Additionally, the maximum likelihood estimation of a constrained version of our model is highly related to another extension of label propagation algorithm, namely, the label propagation algorithm under constraint. Experiments show that the proposed Gaussian stochastic blockmodel performs well on various benchmark networks.

Schrodinger Equation cannot Describe Transition Processes

In this paper we prove the following: (1) The basic error of time-dependent perturbation theory is using the sum of first finite order of perturbed solutions to substitute the exact solution in the divergent interval of the series for calculating the transition probability. In addition quantum mechanics neglects the influence of the normality condition in the continuous case. In both cases Fermi golden rule is not a mathematically reasonable deductive inference from the Schrodinger equation. (2) The transition probability per unit time deduced from the exact solution of the Schrodinger equation is zero, which cannot be used to describe the transition processes.

Answer of the most important unsettled question of physical theory in 20th century

Quantum mechanics take the sum of first finite order approximate solutions of time-dependent perturbation to substitute the exact solution. From the point of mathematics, it may be correct only in the convergent region of the time-dependent perturbation series. Where is the convergent region of this series? Quantum mechanics did not answer this problem. However it is relative to the question, can we use the Schrodinger equation to describe the transition processes? So it is the most important unsettling problem of physical theory. We find out the time-dependent approximate solution for arbitrary and the exact solution. Then we can prove that: (1) In the neighborhood of the conservation of energy, the series is divergent. The basic error of quantum mechanics is using the sum of the first finite orders approximate solutions to substitute the exact solution in this divergent region. It leads to an infinite error. So the Fermi golden rule is not a mathematically reasonable inference of the. Schrodinger equation (2) The transiton probability per unit time deduced from the exact solution of Schrodinger equation cannot describe the transition processes. This paper is only a prime discussion.

GP-B experiment will prove that space-time is flat

Can we obtain the predictive value of GP-B experiment direct from the well known experimental results? This predictive value is more reliable then that deduced from special model of theory. In this paper, we calculate in this way. The result is same as that from special relativistic gravitational theory. So it is extremely likely that GP-B experiment will prove space-time is flat.

Is space-time curved

Whether the space-time is curved or not? The experimental criterions to judge this point are: (1) The results of three classical relativistic experiments in essence are favorable to the special relativistic gravitational theory (base in the flat space-time). However they are unfavorable to the general relativity. (2) In the Gravity Probe-B experiment: the gyroscope precession rate of the orbital effect deduced from the special relativistic gravitational theory =(2/3)* the precession rate of geodetic effect deduced from the general relativity, the precession rate of the earth rotation effect deduced from the special relativistic gravitational theory =(3/2)* the square of cos(phi)* the precession rate of the frame-dragging effect deduced from the general relativity, where (phi) is the angle between the projection of gyroscope angular velocity in the equatorial plane and the normal line of orbital plane. If the experimental values are identical with the predictive values deduced from the special relativistic gravitational theory, then the space-time is flat.