3D point cloud segmentation is one of the key steps in point cloud processing, which is the technology and process of dividing the point cloud data set into several specific regions with unique properties and proposing interesting targets. It has important applications in medical image processing, industrial inspection, cultural relic's identification and 3D visualization. Despite widespread use, point cloud segmentation still faces many challenges because of uneven sampling density, high redundancy, and lack explicit structure of point cloud data. The main goal of this paper is to analyse the most popular algorithms and methodologies to segment point clouds. To facilitate analysis and summary, according to the principle of segmentation we divide the 3D point cloud segmentation methods into edge-based methods, region-based methods, graph-based methods, model-based methods, and machine learning-based methods. Then analyze and discuss the advantages, disadvantages and application scenarios of these segmentation methods. For some algorithms the results of the segmentation and classification is shown. Finally, we outline the issues that need to be addressed and important future research directions.

This paper describes MagicPai's system for Se-mEval 2021 Task 7, HaHackathon: Detecting and Rating Humor and Offense. This task aims to detect whether the text is humorous and how humorous it is. There are four subtasks in the competition. In this paper, we mainly present our solution, a multi-task learning model based on adversarial examples, for task 1a and 1b. More specifically, we first vectorize the cleaned dataset and add the perturbation to obtain more robust embedding representations. We then correct the loss via the confidence level. Finally, we perform interactive joint learning on multiple tasks to capture the relationship between whether the text is humorous and how humorous it is. The final result shows the effectiveness of our system.

No abstract

This paper describes MagicPai's system for Se-mEval 2021 Task 7, HaHackathon: Detecting and Rating Humor and Offense. This task aims to detect whether the text is humorous and how humorous it is. There are four subtasks in the competition. In this paper, we mainly present our solution, a multi-task learning model based on adversarial examples, for task 1a and 1b. More specifically, we first vectorize the cleaned dataset and add the perturbation to obtain more robust embedding representations. We then correct the loss via the confidence level. Finally, we perform interactive joint learning on multiple tasks to capture the relationship between whether the text is humorous and how humorous it is. The final result shows the effectiveness of our system.

On the basis of studying the calibration method of the camera and the rotating platform, we designed an automatic camera calibration scheme based on four azimuth circles, which can realize the automatic sorting of the mark points. In order to improve the efficiency of identifying the center of the calibration plate, this paper uses RANSAC to improve the RED ellipse center detection algorithm. Experimental verification shows that the improved RED algorithm has increased 29.34% in anti-noise interference ability and 30.10% in center detection accuracy, which effectively guarantees the measurement stability and accuracy of the measurement system. Then, with the aid of the designed calibration board, we fit the rotation center of the turntable using the principle of three points in a circle, which provides a basis for the subsequent point cloud splicing. Experiments show that the back-projection error of the calibration method in this paper is less than 1 pixel, and the calibration accuracy is better than Zhang's algorithm.