Search

Semantic image segmentation has been used to detect objects and label pixels in images. It has been applied to high-resolution remote sensing images to detect different types of terrains and landforms. However, the accuracy of the existing methods is not always satisfactory. Here we propose a semantic segmentation post-processing method using K-mean clustering. Our method aggregates the predictions from network training algorithms such as Unet and HrNet [1], and then performs postprocessing using K-Mean clustering iteratively [2] [3]. The accuracy of our method improves as the number of iterations increases. Source code is at https://github.com/carlsummer/SSK.

Pedestrian detection in a crowd is a very challenging issue. This paper addresses this problem by a novel NonMaximum Suppression (NMS) algorithm to better refine the bounding boxes given by detectors. The contributions are threefold: (1) we propose strict-NMS, which applies a higher suppression threshold to an instance; (2) we apply Mask boot module to the Fast-RCNN to improve the detect capacity. (3) We achieve state of the art results on the CityPersons benchmarks.

In order to improve the professional skills of business English talents in our country, the reconstruction method of business English knowledge based on information teaching is proposed. Through the investigation and analysis of the current business English teaching content and knowledge system in our country, the distribution of basic skills such as listening, speaking, reading, writing and translation in business English teaching is rearranged in a targeted way, Improving the teaching content of business English, optimizing the teaching mode of business English, and finally realizing the reconstruction of the knowledge system of business English. Finally, through empirical investigation, it is proved that business English knowledge reconstruction method based on information teaching can better improve students' mastery of professional knowledge and application of professional skills compared with traditional English teaching methods.

This paper has studied the conventional, new and compound methods of acid zing fluid accentuation. The expression of density and concentration of confecting acidizing fluid was gained by experimental data fitting. The expression about density and volume of growth rate of calcium chloride or calcium bromide solution correlating with amount of calcium chloride or calcium bromide was formed. The new acidizing fluid accentuation product was developed, in which the density reached 1.93 gram per cubic centimeter, and it is well compatible with other adjuvant. Complex calculation method was got by studying compounded accentuation technology, which could make the density of acidizing fluid range from 1.1 to 1.9 gram per cubic centimeter according to different hydrochloric acid concentration requirements. The results of these studies could supply technical support and reserves for deep well or ultra deep well acid, and promote oil and gas exploration and development toward the deep well and ultra deep well.

In this paper, we discuss the knowledge redundancy and the knowledge simplification of an uncertain system based on theoretical aspects of rough sets and information entropy. The consistence of analysis on knowledge redundancy and the knowledge simplification is set up. The relationship of information entropy and knowledge significance is drawn on the information dependence of condition attributes and decision condition attributes. A new approach of analysis on knowledge redundancy and the knowledge simplification based on information entropy is presented. The theoretical analysis and the proposed method are validated by some of examples here.

In this paper, a new approach of reducing redundancy condition is put forward based on an information consistency relationship of equivalent classification. The best coverage of data coordinated about the decision-table of an information decision system and the significance of attribute to the system is chosen as a rule of heuristic information of the attribute j by way of ant colony optimization. An algorithm of attribute reduction based on ant colony optimization is suggested to solve N-P hard problem in derivation of a minimal set of attribute reduction. The merit of the optimization of attribute reduction on ant colony algorithm is validated by the proposed application here.

The main process in hydroponics plants cultivating is aquatic induction. Aquatic induction is the process of cultivating plants in the right artificial environment to induce aerenchyma formation, which can make plants be grown in water. With long-time practices and researches, rules and environment parameters for aquatic induction was summarized. And a set of fuzzy control system for hydroponics cultivating was developed. Making use of the experience from practices, a method was put forward to control dissolved oxygen, CO2 and EC in some hydroponics cultivating greenhouses without oxygen detector, CO2 detector and EC detector. The result of the system running shows that the system can achieve hydroponics cultivating without watchers.