This study presents the analogical assessment of the train-induced vibration and radiated noise in a proposed theater. The theater is to be constructed in a region with crowded metro lines, and the assessment is implemented in an analogical building with comparable structural type and metro condition. Prior to the assessment, the comparability of the analogical building with the theater is validated using the train-induced ground vibration. With the same horizontal distance from the metro line, the train-induced vibration level in the analogical building is 9 dB higher than that in the construction site of the theater. Such results indicate that the lack of soil layers may lead to a dramatic increase in train-induced vibration in the building. In the staircase of the analogical building, the train-induced radiated noise reached 55 dB (A), which is 10 dB (A) higher than the daytime allowable level. As the most important indicator, the noise rating number in the cinema of the analogical building is NR-43, which put forward an enormous challenge on the construction of the theater with a denoise demand of 23 dB. The analogical method applied in this study provides an effective and practical way for the assessment of train-induced vibration and radiated noise in proposed vibration-sensitive buildings. The assessment results that provide necessary reference and support for the anti-vibration design will help guarantee the stage effect of the theater.@en

To address the problem of training on small datasets for action recognition tasks, most prior works are either based on a large number of training samples or require pre-trained models transferred from other large datasets to tackle overfitting problems. However, it limits the research within organizations that have strong computational abilities. In this work, we try to propose a data-efficient framework that can train the model from scratch on small datasets while achieving promising results. Specifically, by introducing a 3D central difference convolution operation, we proposed a novel C3D neural network-based two-stream (Rank Pooling RGB and Optical Flow) framework for the task. The method is validated on the action recognition track of the ECCV 2020 VIPriors challenges and got the 2nd place (88.31%). It is proved that our method can achieve a promising result even without a pre-trained model on large scale datasets. The code will be released soon. @en